feat: auto-generate Gemini CLI extension from Claude plugins - #369
Merged
openshift-merge-bot[bot] merged 1 commit intoMar 13, 2026
Merged
openshift-merge-bot[bot] merged 1 commit into
openshift-merge-bot[bot] merged 1 commit into
Conversation
Contributor
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
WalkthroughThis PR adds Gemini CLI support to the project by introducing a conversion script that transforms Claude Code plugins into Gemini CLI extensions. It updates project documentation to describe dual compatibility with Claude Code and Gemini CLI, adds a build target for running the conversion, and configures code review to exclude generated files. Changes
Sequence DiagramsequenceDiagram
participant User
participant Script as convert_to_gemini.py
participant FS as File System
participant Manifest as gemini-extension.json
participant Context as GEMINI.md
User->>Script: Run conversion (--plugin or default)
Script->>FS: Discover plugins/ directories
FS-->>Script: Return plugin list
loop For each plugin
Script->>FS: Read plugin.json & commands/
FS-->>Script: Plugin metadata & MD files
Script->>Script: Convert MD to TOML (extract frontmatter)
Script->>FS: Write commands/{plugin}/*.toml
Script->>FS: Read skills/ from plugin
FS-->>Script: Skill directories
Script->>FS: Copy to skills/{plugin}-{skill}/
end
Script->>Script: Determine version bump needed
Script->>Manifest: Generate/update manifest
Manifest-->>Script: Bumped version
Script->>Context: Generate context file
Context-->>Script: GEMINI.md created
Script-->>User: Conversion complete with summary
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~30 minutes Important Pre-merge checks failedPlease resolve all errors before merging. Addressing warnings is optional. ❌ Failed checks (1 inconclusive)
✅ Passed checks (7 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
📝 Coding Plan
Comment Tip CodeRabbit can use your project's `ruff` configuration to improve the quality of Python code reviews.Add a Ruff configuration file to your project to customize how CodeRabbit runs |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
Note
Due to the large number of review comments, Critical severity comments were prioritized as inline comments.
🟠 Major comments (41)
commands/git/fix-cherrypick-robot-pr.toml-57-74 (1)
57-74:⚠️ Potential issue | 🟠 MajorGit remote discovery logic is flawed.
git config user.namereturns the human-readable display name (e.g., "John Doe"), not the GitHub username. GitHub remote URLs use usernames (e.g.,github.com/johndoe/repo), so the grep patterns on lines 60 and 63 will almost never match actual remote URLs.This means the "auto-discovery" will silently fail and always fall back to the hardcoded defaults (
origin/upstream), making the discovery logic misleading.Consider using
gh api user --jq .loginto get the actual GitHub username, or parsing the remote URL directly.🔧 Proposed fix for remote discovery
# Discover the upstream remote (the main repository) -# Look for a remote that's not owned by the current user -UPSTREAM_REMOTE=$(git remote -v | grep "fetch" | grep -v "$(git config user.name)" | awk '{print $1}' | head -1) +# Look for a remote that's not owned by the current GitHub user +GH_USER=$(gh api user --jq .login 2>/dev/null || echo "") +if [ -n "$GH_USER" ]; then + UPSTREAM_REMOTE=$(git remote -v | grep "fetch" | grep -v "$GH_USER" | awk '{print $1}' | head -1) + FORK_REMOTE=$(git remote -v | grep "$GH_USER.*push" | awk '{print $1}' | head -1) +else + # Fallback if gh CLI not authenticated + UPSTREAM_REMOTE="" + FORK_REMOTE="" +fi -# Discover the fork remote (your fork) -FORK_REMOTE=$(git remote -v | grep "$(git config user.name).*push" | awk '{print $1}' | head -1) # If not found, fall back to common names UPSTREAM_REMOTE=${UPSTREAM_REMOTE:-upstream} FORK_REMOTE=${FORK_REMOTE:-origin}Note: The same issue exists at line 128 where the pattern is repeated.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/fix-cherrypick-robot-pr.toml` around lines 57 - 74, The remote discovery uses `git config user.name` (human display name) to grep remotes, which won't match GitHub usernames; update the logic that sets `UPSTREAM_REMOTE` and `FORK_REMOTE` to derive the GitHub username from `gh api user --jq .login` or parse each remote URL (via `git remote get-url <name>`) to extract the owner segment and match against that owner, then select the non-fork remote as UPSTREAM_REMOTE and the matching-owner remote as FORK_REMOTE; apply the same fix to the repeated pattern where `UPSTREAM_REMOTE`/`FORK_REMOTE` are determined later in the file.commands/git/redescribe.toml-88-96 (1)
88-96:⚠️ Potential issue | 🟠 MajorUse
--body-fileor stdin to pass multiline Markdown instead of inline--body.The inline
--bodyapproach is fragile: quotes, backticks, dollar signs, and newlines in the generated description will corrupt the payload. GitHub CLI'sgh pr editsupports the-F, --body-fileoption to safely handle multiline content:Safe approaches:
# From a file gh pr edit <pr-url> --body-file pr-body.md# From stdin gh pr edit <pr-url> --body-file - <<'EOF' <new-description> EOF🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/redescribe.toml` around lines 88 - 96, Replace the fragile inline gh command that uses `gh pr edit <pr-url> --body "<new-description>"` with a safe `--body-file` approach: write the generated multiline Markdown into a temporary file (or pipe it to stdin) and call `gh pr edit <pr-url> --body-file <file-or->` instead of `--body`, ensuring you use the same `gh pr edit` invocation and remove the inline `--body` usage to avoid quoting/newline issues.commands/node-tuning/generate-tuned-profile.toml-76-76 (1)
76-76:⚠️ Potential issue | 🟠 MajorFix the invalid
oc applydry-run flag.Line 76 uses
oc apply --server-dry-run=client, which is not a valid flag combination. The correct syntax uses--dry-run=clientor--dry-run=serverinstead.Suggested fix
- - Optionally run `oc apply --server-dry-run=client -f .work/node-tuning/$PROFILE_NAME/tuned.yaml` to confirm schema compatibility. + - Optionally run `oc apply --dry-run=client -f .work/node-tuning/$PROFILE_NAME/tuned.yaml` to confirm local validation. + - If you want API-server validation, use `oc apply --dry-run=server -f .work/node-tuning/$PROFILE_NAME/tuned.yaml`.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/node-tuning/generate-tuned-profile.toml` at line 76, The README/command text uses an invalid oc flag combination: replace the incorrect string "oc apply --server-dry-run=client -f .work/node-tuning/$PROFILE_NAME/tuned.yaml" with the correct dry-run form, e.g. "oc apply --dry-run=client -f .work/node-tuning/$PROFILE_NAME/tuned.yaml" (or use "--dry-run=server" if server-side dry-run is intended) so the instruction in generate-tuned-profile.toml is valid.commands/node/cluster-node-health-check.toml-60-63 (1)
60-63:⚠️ Potential issue | 🟠 Major
--output-format jsonis documented but not implemented.
OUTPUT_FORMATis initialized, but every execution path still prints text and exits; the JSON block here is only an example. As written,/node:cluster-node-health-check --output-format jsoncannot produce the machine-readable output the manifest promises.Also applies to: 482-535
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/node/cluster-node-health-check.toml` around lines 60 - 63, The manifest documents an --output-format flag but the code never emits JSON: update the command handling that reads OUTPUT_FORMAT (the --output-format flag) to branch on its value and produce machine-readable JSON when OUTPUT_FORMAT == "json" instead of always printing text; locate the main execution/print logic that formats the health results (the block that currently prints human-readable status lines) and add a JSON serialization path that builds a structured object (statuses, node info, timestamps, exit code) and writes it to stdout, ensuring the text output remains the default when OUTPUT_FORMAT is absent or set to "text".commands/hcp/cluster-health-check.toml-7-11 (1)
7-11:⚠️ Potential issue | 🟠 MajorAdd the optional namespace positional to the synopsis.
The implementation and examples accept a second positional namespace argument, but the synopsis omits it. That makes the documented two-positional form look invalid.
Also applies to: 49-53
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/hcp/cluster-health-check.toml` around lines 7 - 11, The Synopsis block currently lists only one positional (<cluster-name>) but the implementation and examples accept an optional second positional namespace; update the synopsis string `/hcp:cluster-health-check <cluster-name> [--verbose] [--output-format json|text]` to include the optional namespace (e.g. `/hcp:cluster-health-check <cluster-name> [<namespace>] [--verbose] [--output-format json|text]`) and make the same change to the other Synopsis occurrence referenced in the file so the documented form matches the implemented command signature.commands/hcp/cluster-health-check.toml-58-60 (1)
58-60:⚠️ Potential issue | 🟠 Major
--output-format jsoncurrently has no execution path.The prompt advertises a JSON mode, but the implementation never branches on that flag and only shows a sample payload. Consumers expecting machine-readable output will still get text.
Also applies to: 472-533
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/hcp/cluster-health-check.toml` around lines 58 - 60, The CLI advertises a --output-format flag but no code path uses it; update the main command handler (the function that processes the cluster-health-check command, e.g., the handler/execute function for this subcommand) to read the --output-format value and branch: if value === "json" then serialize the health-check result object to JSON and write it to stdout (instead of the human-readable renderer), otherwise use the existing text renderer; ensure error conditions still return appropriate exit codes and that the JSON output contains the same structured fields shown in the sample payload; add/update unit/integration tests for both "text" and "json" modes to validate output.commands/lvms/analyze.toml-159-172 (1)
159-172:⚠️ Potential issue | 🟠 MajorThe fallback must-gather reader drops old-namespace support.
Earlier sections promise compatibility with both
openshift-lvm-storageandopenshift-storage, but the built-in file-reading path hardcodes onlyopenshift-lvm-storage. If the Python helper is missing, older must-gathers will fail despite the documented backward-compatibility claim.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/lvms/analyze.toml` around lines 159 - 172, The fallback built-in file-reading commands hardcode the namespace "openshift-lvm-storage", dropping support for older "openshift-storage" must-gathers; update the commands that read lvmclusters.yaml, lvmvolumegroups.yaml and the pods.yaml/events.yaml lines so they try both namespaces (e.g., use find with -name "lvmclusters.yaml" and -name "lvmvolumegroups.yaml" as before but search both namespace paths, or change the cat lines to cat {must-gather-path}/namespaces/{openshift-lvm-storage,openshift-storage}/pods.yaml and cat {must-gather-path}/namespaces/{openshift-lvm-storage,openshift-storage}/events.yaml), ensuring the fallback supports both "openshift-lvm-storage" and "openshift-storage" so older must-gathers are handled as documented.commands/gwapi/install.toml-83-123 (1)
83-123:⚠️ Potential issue | 🟠 MajorThe polling snippet never fails on timeout.
Both loops stop when
ELAPSED == TIMEOUT, but there is no post-loop check before continuing. A permanently unready GatewayClass/Gateway can therefore fall through into the final summary as if installation completed.Suggested fix
TIMEOUT=300 INTERVAL=5 ELAPSED=0 + GATEWAYCLASS_READY=false + GATEWAY_READY=false # Wait for GatewayClass while [ $ELAPSED -lt $TIMEOUT ]; do ACCEPTED=$(oc get gatewayclass <name> -o jsonpath='{.status.conditions[?(@.type=="Accepted")].status}' 2>/dev/null) if [ "$ACCEPTED" = "True" ]; then echo "✓ GatewayClass is accepted" + GATEWAYCLASS_READY=true break fi echo "Waiting for GatewayClass to be accepted... ($(($ELAPSED))s / ${TIMEOUT}s)" sleep $INTERVAL ELAPSED=$(($ELAPSED + $INTERVAL)) done + + if [ "$GATEWAYCLASS_READY" != "true" ]; then + echo "Timeout waiting for GatewayClass to be accepted. Current status:" + oc get gatewayclass <name> -o yaml + exit 1 + fi # Wait for Gateway ELAPSED=0 while [ $ELAPSED -lt $TIMEOUT ]; do PROGRAMMED=$(oc get gateway <name> -n <namespace> -o jsonpath='{.status.conditions[?(@.type=="Programmed")].status}' 2>/dev/null) ACCEPTED=$(oc get gateway <name> -n <namespace> -o jsonpath='{.status.conditions[?(@.type=="Accepted")].status}' 2>/dev/null) if [ "$PROGRAMMED" = "True" ] && [ "$ACCEPTED" = "True" ]; then echo "✓ Gateway is ready" + GATEWAY_READY=true break fi echo "Waiting for Gateway to be ready... ($(($ELAPSED))s / ${TIMEOUT}s)" sleep $INTERVAL ELAPSED=$(($ELAPSED + $INTERVAL)) done + + if [ "$GATEWAY_READY" != "true" ]; then + echo "Timeout waiting for Gateway to be ready. Current status:" + oc get gateway <name> -n <namespace> -o yaml + exit 1 + fi🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/gwapi/install.toml` around lines 83 - 123, The polling loops using TIMEOUT/INTERVAL/ELAPSED for GatewayClass and Gateway do not check if they exited due to timeout; after each while loop (for GatewayClass checks and for Gateway checks) add an explicit check of whether the condition was satisfied (e.g., test that ACCEPTED/PROGRAMMED variables are True), and if not then print "Timeout waiting for resources to be ready. Current status:", dump full resource status with oc get gatewayclass <name> -o yaml and oc get gateway <name> -n <namespace> -o yaml, and exit with a non-zero status; ensure you reference the same variables (ELAPSED, TIMEOUT, ACCEPTED, PROGRAMMED) and commands (oc get gatewayclass, oc get gateway) so the timeout branch runs when the loop ends without success.commands/node/cluster-node-health-check.toml-33-50 (1)
33-50:⚠️ Potential issue | 🟠 MajorAdd
jqas a documented prerequisite.The implementation extensively uses
jqfor JSON processing (extracting node conditions, pod statuses, resource capacities, labels, and more), but the Prerequisites section (lines 33-50) only mentionsoc/kubectl. The command will fail immediately when attempting the firstjqpipeline ifjqis not installed. Addjqto the prerequisites with installation instructions.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/node/cluster-node-health-check.toml` around lines 33 - 50, Add jq to the prerequisites for the cluster-node-health-check command: update the Prerequisites section to list jq as a required tool, include installation links for common platforms (e.g., package managers or official site) and a verification command such as `jq --version`, and mention that the script relies on jq for JSON processing of node conditions, pods, capacities and labels so the command will fail without it.commands/git/backport.toml-57-63 (1)
57-63:⚠️ Potential issue | 🟠 MajorClarify the sequence for conflict resolution and branch transitions.
The workflow tells the user to run
git cherry-pick --continueorgit cherry-pick --abort, then asks about continuing to the next branch. However, it doesn't explicitly state that the user must complete one of those commands (resolving the cherry-pick state) before the system can proceed to checkout the next branch. Without this clarity, an implementer might incorrectly attempt to switch branches while the cherry-pick is still in progress, causing the workflow to fail.Revise lines 61-62 to make the sequence explicit: the user must first complete the conflict resolution command, and only after that is resolved will the next branch be checked out.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/backport.toml` around lines 57 - 63, Make the conflict-resolution wording explicit: update the block that mentions `git cherry-pick --continue` / `git cherry-pick --abort` so it states that the user must finish the cherry-pick state (by running `git cherry-pick --continue` after resolving conflicts or `git cherry-pick --abort`) before the tool or user can switch branches; only after the cherry-pick is completed should the workflow attempt to checkout the next branch or prompt “continue to the next branch.” Reference the existing phrases `git status`, `git diff`, `git cherry-pick --continue`, `git cherry-pick --abort`, and “checkout the next branch” when updating the text to enforce this sequence.commands/gwapi/install.toml-39-42 (1)
39-42:⚠️ Potential issue | 🟠 MajorThis manifest overstates Kubernetes support—it's functionally OpenShift-only.
The introduction claims Kubernetes/OpenShift compatibility and "uses
oc(preferred) orkubectl", but multiple critical steps hardcodeoc:
- Line 40: Domain retrieval uses
oc get ingresses.config/cluster(OpenShift-specific API; Kubernetes has no equivalent)- Line 61: Domain substitution hardcodes
oc applywithout conditional selection- Lines 91–92, 104–105, 122: Readiness checks and diagnostics hardcode
octhroughoutThe fallback at lines 41–42 (ask user for manual domain entry) is incomplete: even if a user supplies the domain, the workflow fails on pure Kubernetes at every subsequent step. Either remove Kubernetes from the supported platforms or implement conditional CLI detection and use throughout.
Also applies to: 61, 91–92, 104–105, 122, 194
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/gwapi/install.toml` around lines 39 - 42, The manifest currently hardcodes OpenShift 'oc' usage (e.g., DOMAIN=$(oc get ingresses.config/cluster -o jsonpath={.spec.domain}), all oc apply/oc get readiness checks and diagnostics), so either remove Kubernetes from supported platforms or implement conditional CLI/cluster detection: detect whether 'oc' is available and cluster is OpenShift, else fall back to 'kubectl' and use Kubernetes-appropriate ways to determine the external domain (e.g., examine Ingress/IngressClass, Service LoadBalancer e.g., EXTERNAL-IP, or prompt with clear instructions), and apply that conditional logic consistently for the DOMAIN retrieval, the resource application steps (currently using oc apply), readiness checks/diagnostics (currently using oc get/oc logs), and every other oc occurrence noted (lines referenced in the review). Update messages and flow so a user-provided DOMAIN will work on both paths and ensure all commands reference the chosen CLI variable instead of hardcoded 'oc'.commands/ci/query-job-status.toml-7-10 (1)
7-10:⚠️ Potential issue | 🟠 MajorKeep the
ci:namespace in every documented invocation.This file declares
ci:query-job-statuson Line 5, but the synopsis and all examples document/query-job-statuswithout the namespace. Users following this prompt will call a nonexistent command.🛠️ Suggested change
-/query-job-status <execution-id> +/ci:query-job-status <execution-id> ... -The `query-job-status` command queries the status of a gangway job execution via the REST API using the execution ID returned when a job is triggered. +The `ci:query-job-status` command queries the status of a gangway job execution via the REST API using the execution ID returned when a job is triggered. ... - /query-job-status ca249d50-dee8-4424-a0a7-6dd9d5605267 + /ci:query-job-status ca249d50-dee8-4424-a0a7-6dd9d5605267 ... - /query-job-status 8f3a9b2c-1234-5678-9abc-def012345678 + /ci:query-job-status 8f3a9b2c-1234-5678-9abc-def012345678 ... - /query-job-status 5a6b7c8d-9e0f-1a2b-3c4d-5e6f7a8b9c0d + /ci:query-job-status 5a6b7c8d-9e0f-1a2b-3c4d-5e6f7a8b9c0dAlso applies to: 14-19, 63-66, 78-81, 84-87
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/query-job-status.toml` around lines 7 - 10, The synopsis and examples omit the required namespace; update every user-facing invocation to include the declared command name ci:query-job-status (e.g., change occurrences of `/query-job-status` to `/ci:query-job-status`) for the synopsis and all example blocks noted (lines around 14-19, 63-66, 78-81, 84-87) so documented commands match the actual declared symbol ci:query-job-status.commands/jira/solve.toml-27-27 (1)
27-27:⚠️ Potential issue | 🟠 MajorFix JIRA issue URL parameter expansion.
Lines 27 and 36 use
{$1}in the REST API URL, which leaves literal braces in the path. This produces/issue/{OCPBUGS-12345}instead of/issue/OCPBUGS-12345, breaking the API request. Use${1}for correct shell parameter expansion.Changes required
- The command uses curl to fetch JIRA data via REST API: https://issues.redhat.com/rest/api/2/issue/{$1} + The command uses curl to fetch JIRA data via REST API: https://issues.redhat.com/rest/api/2/issue/${1}- Use curl to fetch JIRA issue data: curl -s "https://issues.redhat.com/rest/api/2/issue/{$1}" + - Use curl to fetch JIRA issue data: curl -s "https://issues.redhat.com/rest/api/2/issue/${1}"Also applies to: 36-36
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/solve.toml` at line 27, The JIRA REST API URL in the curl invocations currently uses the literal brace form "https://issues.redhat.com/rest/api/2/issue/{$1}" which leaves braces in the path; update both occurrences of that URL (the curl command strings referencing {$1}) to use shell parameter expansion "${1}" instead of "{$1}" so the request becomes "/issue/OCPBUGS-12345" at runtime.commands/ci/query-job-status.toml-52-59 (1)
52-59:⚠️ Potential issue | 🟠 MajorRemove Claude-specific instructions and Skill tool references from this file.
Lines 52-59 and 82-88 contain instructions explicitly labeled "Important for Claude" that reference the "Skill tool" (a Claude-specific concept) and describe Claude-specific behaviors in the examples. These conflict with the Gemini conversion goal and will mislead Gemini users. Replace with tool-agnostic instructions that describe what the command does (parse JSON, display status, show artifact paths) without mentioning Claude or the Skill tool.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/query-job-status.toml` around lines 52 - 59, Remove the "Important for Claude" block and any references to the "Skill tool" and Claude-specific instructions; instead replace those paragraphs (the ones that mention using the Skill tool and Claude) with a tool-agnostic description that (1) tells the command to ensure authentication is available before running curl_with_token.sh, (2) instructs the runner to locate and verify curl_with_token.sh is present in the repo before executing it, and (3) specifies the expected behavior: parse the command's JSON response, present it in a readable format, prominently display the job status, note that PENDING/RUNNING means the job is still in progress and SUCCESS/FAILURE indicates completion, and if a gcs_path is present include the artifact path.commands/jira/issues-by-component.toml-45-49 (1)
45-49:⚠️ Potential issue | 🟠 MajorThis prompt still targets the Claude plugin runtime instead of the generated Gemini extension.
Lines 45-49 describe Claude-specific limits, and Lines 194-198 hardcode
plugins/jira/skills/jira-issues-by-component/jira_curl.sh. In this PR the generated extension ships commands and skills in the new root layout, so these instructions now point at the wrong runtime and a nonexistent wrapper path. Please fix the converter too, or regeneration will reintroduce this.Also applies to: 193-198
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/issues-by-component.toml` around lines 45 - 49, Update the prompt and converter so it targets the generated Gemini extension and the new root layout instead of the Claude plugin runtime: replace any Claude-specific language (e.g., references to Claude limits in commands/jira/issues-by-component.toml) and change hardcoded wrapper paths like "plugins/jira/skills/jira-issues-by-component/jira_curl.sh" to the new extension/skill command locations produced by the generator (the root-level commands/skills layout used by the generated Gemini extension); update the converter logic that emits these strings so future regeneration produces the correct runtime-targeted prompt and the correct root-level command paths.commands/jira/clone-from-github.toml-91-123 (1)
91-123:⚠️ Potential issue | 🟠 MajorProfile selection is described with three different identifiers.
These sections talk about profile “name” values like
my-profileanddefault, but the documentedprofiles.yamlonly exposesdescription, and Line 287 then tells users to pass"OLM Project". The setup and usage docs need one canonical profile identifier or users will call the command with values that cannot be resolved.Also applies to: 272-287
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/clone-from-github.toml` around lines 91 - 123, The docs show inconsistent profile identifiers (flags `--profile` vs `--profile-name`, examples using `my-profile`/`default`, and elsewhere referencing human labels like "OLM Project") which will confuse users because `profiles.yaml` exposes only keys like `description`; normalize to a single canonical identifier: pick one flag name (e.g., `--profile-name`) and update all references (examples, flag mapping, and the section describing profiles) to state that the CLI expects the profile key from profiles.yaml (not the description/human label), and change the example that uses `"OLM Project"` to use the actual profile key that appears in profiles.yaml so the documented usage and the profiles.yaml schema match.commands/ci/trigger-periodic.toml-5-10 (1)
5-10:⚠️ Potential issue | 🟠 MajorUse the installed
/ci:command names in the syntax and examples.Line 5 defines
ci:trigger-periodic, but the synopsis, follow-up suggestion, and examples all publish/trigger-periodicand/query-job-status. In the generated extension those commands are namespaced, so the current docs tell users to invoke the wrong command names.Also applies to: 89-89, 104-122
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/trigger-periodic.toml` around lines 5 - 10, The docs use the un-namespaced commands (/trigger-periodic, /query-job-status) but the installed extension exposes namespaced commands (ci:trigger-periodic); update the synopsis, examples and any follow-up mentions to use the installed command names (e.g. /ci:trigger-periodic and /ci:query-job-status) so users invoke the correct names; search for occurrences of "/trigger-periodic" and "/query-job-status" (including the blocks around ci:trigger-periodic and the sections at lines ~89 and ~104-122) and replace them with the namespaced equivalents while keeping the same argument and ENV_VAR examples.commands/jira/issues-by-component.toml-124-151 (1)
124-151:⚠️ Potential issue | 🟠 MajorMap friendly time windows to valid JQL before composing the query.
Lines 125-128 accept values like
last-weekandYYYY-MM-DD:YYYY-MM-DD, but Lines 148-151 interpolate that token directly ascreated >= -{time-period}. That yields invalid JQL for the named presets and for custom ranges. This needs an explicit translation step first, e.g.last-week→-7d, and custom ranges → bounded>=/<=clauses.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/issues-by-component.toml` around lines 124 - 151, The time-period token is being interpolated directly into the JQL ("created >= -{time-period}"), which produces invalid JQL for named presets and custom ranges; add a translation step that maps the incoming time-period values (e.g., "last-week", "last-2-weeks", "last-month") to JQL relative durations ("-7d", "-14d", "-30d") and converts custom ranges "YYYY-MM-DD:YYYY-MM-DD" into explicit bounded clauses ("created >= YYYY-MM-DD AND created <= YYYY-MM-DD") before composing the query, then use the resulting JQL fragment instead of the raw token when building the base query.commands/must-gather/windows.toml-62-66 (1)
62-66:⚠️ Potential issue | 🟠 MajorThis execution step still points at the old plugin-local analyzer path.
It invokes
plugins/must-gather/skills/must-gather-analyzer/scripts/analyze_windows_logs.py, which will not exist inside the generated Gemini extension. As written, the first real action for the command is a broken path.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/must-gather/windows.toml` around lines 62 - 66, Update the execution step in commands/must-gather/windows.toml so it no longer references the removed plugin-local path "plugins/must-gather/skills/must-gather-analyzer/scripts/analyze_windows_logs.py"; replace it with the analyzer entry point that will exist inside the generated Gemini extension (e.g., the extension-relative script or CLI installed with the extension) so the command calls the correct analyze_windows_logs entry in the extension package rather than the old plugin path.commands/ci/trigger-periodic.toml-26-42 (1)
26-42:⚠️ Potential issue | 🟠 MajorThe security section still addresses Claude instead of Gemini.
These sections use “Claude MUST” / “Important for Claude” language even though this file is part of the generated Gemini extension. That leaves the runtime guidance ambiguous and misses the conversion goal for this PR. Please fix the generator so the platform-specific wording is rewritten everywhere.
Also applies to: 95-100
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/trigger-periodic.toml` around lines 26 - 42, Update the platform-specific wording in the security/confirmation section so all occurrences of the assistant name refer to Gemini instead of Claude: search for phrases like "Claude MUST", "Important for Claude", and any lines in the "MANDATORY USER CONFIRMATION" block (and the similar block around lines 95-100) and replace them with "Gemini MUST" / "Important for Gemini" (or equivalent generator variables if this is templated) so the generated Gemini extension uses the correct assistant name consistently; ensure the change is applied to every instance in the file and in the generator template so no residual "Claude" references remain.commands/hcp/generate.toml-349-354 (1)
349-354:⚠️ Potential issue | 🟠 MajorThe skills reference still points at the source plugin tree.
These instructions tell users to inspect
plugins/hypershift/skills/..., but those paths do not exist in the generated Gemini extension. The copied skills need to be referenced via the new extension layout, or these lookup steps will fail immediately.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/hcp/generate.toml` around lines 349 - 354, The docs incorrectly point to the original plugin tree (plugins/hypershift/skills/...); update the instructions in generate.toml so they reference the actual generated Gemini extension layout where the skill files were copied (e.g., the extension's skills directory and the SKILL.md files such as hcp-create-aws/SKILL.md and hcp-create-kubevirt/SKILL.md). Replace occurrences of "plugins/hypershift/skills/..." with the correct extension-relative path used by the build output (the extension's skills folder) so the example commands (ls/cat) succeed against the generated package.commands/jira/setup-gh2jira.toml-199-212 (1)
199-212:⚠️ Potential issue | 🟠 MajorRed Hat Jira users are routed through the wrong auth flow here.
This block classifies
issues.redhat.comas Atlassian Cloud, then immediately describes Red Hat Jira as a separate case. That contradicts the other Jira manifests in this PR, which treat Red Hat Jira as its own PAT-based flow, and it will send Red Hat users to the wrong token creation instructions.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/setup-gh2jira.toml` around lines 199 - 212, The block incorrectly treats issues.redhat.com as Atlassian Cloud; update the "For Atlassian Cloud" and "For Red Hat Jira" sections so that issues.redhat.com is routed to the Red Hat PAT-based flow instead of the Atlassian token flow: change the classification logic/text so "issues.redhat.com" is explicitly handled by the Red Hat Jira instructions and move any Red Hat-specific notes (SSO/Kerberos, org docs) into the "For Red Hat Jira" section while keeping the Atlassian Cloud steps for *.atlassian.net only and ensuring the token creation steps reflect the PAT-based flow for Red Hat Jira.commands/golang/lint-fix.toml-22-29 (1)
22-29:⚠️ Potential issue | 🟠 MajorPreserve the user’s lint scope when falling back from
hack/go-lint.sh.Lines 22-25 say extra flags are passed through, but Lines 26-29 replace that with a plain
golangci-lint run --fixwhen the wrapper is used. That fallback drops any package/file scoping the user requested and can end up rewriting a much larger surface than intended.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/golang/lint-fix.toml` around lines 22 - 29, The fallback that runs a plain "golangci-lint run --fix" when "hack/go-lint.sh" is detected drops user-provided package/file scope and flags; update the fallback logic so that the exact extra flags (including package/file targets and any --config=...) captured by the "Go Lint" skill are preserved and appended to the fallback command (i.e., build the command as "golangci-lint run --fix <captured-flags>"), ensuring the wrapper detection for hack/go-lint.sh still chooses direct golangci-lint but retains all original flags and scope.commands/must-gather/ovn-dbs.toml-47-55 (1)
47-55:⚠️ Potential issue | 🟠 MajorThe analyzer lookup still assumes the source plugin directory structure.
These steps tell the agent to find
analyze_ovn_dbs.pyunderplugins/must-gather/...or~/.cursor/commands/ai-helpers/plugins/must-gather. This PR generates a standalone Gemini extension with copied skills in the new extension layout, so that search path will not resolve at runtime.Also applies to: 60-71
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/must-gather/ovn-dbs.toml` around lines 47 - 55, The analyzer lookup in ovn-dbs.toml currently assumes the old plugin directory layout and hardcodes the path to analyze_ovn_dbs.py; update the resolution so it does not rely on "plugins/must-gather/..." layout—either (a) make the TOML reference a logical resource name and change the runtime to locate the script by scanning the plugin/extension installation root for "analyze_ovn_dbs.py" (recursively) or (b) register the script as a package resource/entry point and load it via importlib.resources or pkg_resources; modify the code that reads ovn-dbs.toml to try both the legacy path and the new extension layout discovery (search for the file name analyze_ovn_dbs.py) so the analyzer works for standalone Gemini extensions as well.commands/git/branch-cleanup.toml-137-146 (1)
137-146:⚠️ Potential issue | 🟠 MajorProtected branches are still eligible for deletion in the example logic.
This loop only excludes
$main_branch, sodevelop,release/*, andhotfix/*can still be surfaced as delete candidates even though the safety section says they must never be removed.Also applies to: 271-276
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/branch-cleanup.toml` around lines 137 - 146, The loop that lists merged branches only excludes $main_branch so protected branches like "develop", "release/*", and "hotfix/*" still get output; update the branch-filtering logic (in the loop that iterates over git branch, referencing variables main_branch, check_if_merged, merge_status and method) to skip any branch that matches protected names/patterns — e.g., add a check (or an is_protected_branch function and a protected_branches list) that returns true for "develop" and for patterns "release/*" and "hotfix/*" and continue to next branch when protected so they are never printed as delete candidates.commands/jira/grooming.toml-123-132 (1)
123-132: 🛠️ Refactor suggestion | 🟠 MajorClose the fenced block before
## Output Format.The example that starts at Line 123 never closes its ```bash fence, so everything after it renders as code instead of structured help.
Proposed fix
12. **Combine Label and status filters**: ```bash /jira:grooming OCPBUGS last-week --label "performance" --status "NEW" + ``` ## Output Format🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/grooming.toml` around lines 123 - 132, The fenced code block that begins with ```bash at the /jira:grooming example is not closed, causing subsequent content (including "## Output Format") to render as code; locate the example containing the "/jira:grooming OCPBUGS last-week --label \"performance\" --status \"NEW\"" line (under the "Combine Label and status filters" section) and add the closing ``` fence immediately after that command so the code block is properly terminated and the following "## Output Format" renders as normal text.commands/ci/ask-sippy.toml-4-10 (1)
4-10:⚠️ Potential issue | 🟠 MajorThis converted descriptor still advertises the Claude-era command and agent names.
The file maps to
ci:ask-sippy, but the synopsis/examples still use/ask-sippy, and the operational notes still talk aboutClaude. That leaves the generated Gemini command inconsistent with both its path and the PR’s conversion goal.Also applies to: 29-40, 70-76, 80-100
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/ask-sippy.toml` around lines 4 - 10, The descriptor for the command ci:ask-sippy is inconsistent: update all occurrences of the old Claude-era command and agent names (e.g., the `/ask-sippy` examples and any mentions of "Claude") to match the converted Gemini-style command and agent naming; search for the symbols `ci:ask-sippy`, `/ask-sippy`, and the string `Claude` in this file and replace the synopsis, examples, and operational notes (including the sections noted around lines 29-40, 70-76, 80-100) so the command name, usage examples, and agent references consistently reflect the new Gemini conversion.commands/ci/check-if-jira-regression-is-ongoing.toml-134-139 (1)
134-139:⚠️ Potential issue | 🟠 MajorRecent-run analysis still assumes
test_idexists.The workflow explicitly supports bugs that only expose a test name or regression ID, but this step always calls
fetch-test-runs "$test_id". For those valid inputs, the command dead-ends here instead of degrading gracefully.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/check-if-jira-regression-is-ongoing.toml` around lines 134 - 139, The step unconditionally calls fetch-test-runs (script_path="plugins/ci/skills/fetch-test-runs/fetch_test_runs.py") using the variable test_id even when only a test name or regression ID is provided; update the logic to first verify test_id is set and non-empty before invoking python3 "$script_path" "$test_id" --include-success --format json, and if test_id is absent, branch to a graceful fallback (e.g., call the fetch-test-runs script with an alternative identifier such as test_name or regression_id, or skip this check and mark the result as "not-applicable") so the workflow does not dead-end when only test_name or regression ID inputs exist.commands/ci/check-if-jira-regression-is-ongoing.toml-49-55 (1)
49-55:⚠️ Potential issue | 🟠 MajorUse Jira version metadata before falling back to the latest release.
This prompt never extracts any release field from the Jira payload, so bugs whose summary/description omit
4.xxwill be checked against whateverfetch-releases --latestreturns. That can send the analysis to the wrong stream for older still-supported branches.Also applies to: 72-78
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/check-if-jira-regression-is-ongoing.toml` around lines 49 - 55, The prompt parser currently extracts summary, comments, status, components, and progress.level but never reads Jira release/version metadata, so it falls back to running fetch-releases --latest; update the JSON extraction to prefer Jira's release/version fields (e.g., fields.fixVersions, fields.versions, or a "release" field in the payload) and use that value as the target branch/version before calling fetch-releases --latest; apply the same change to the other parsing block that mirrors this logic (the block referenced in the review as also applying to lines 72-78).commands/gwapi/delete.toml-19-20 (1)
19-20:⚠️ Potential issue | 🟠 MajorNamespace-scoped cleanup should not implicitly delete the shared
GatewayClass.Passing
/gwapi:delete gateway-systemreads like a namespace-local operation, but this prompt still removes the cluster-scopedGatewayClass. If other namespaces use that class, a scoped cleanup becomes a cluster-wide outage.Also applies to: 77-82, 124-128
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/gwapi/delete.toml` around lines 19 - 20, The docs currently state that passing `$1` (optional) for a namespace will delete namespace-scoped Gateway resources and the cluster-scoped `GatewayClass`, which is unsafe; update the description to clarify that providing a namespace (e.g., `gateway-system`) only removes resources in that namespace (from `openshift-ingress` YAMLs if relevant) and does NOT delete the cluster-scoped `GatewayClass` unless an explicit cluster-scoped flag or no-namespace invocation is used; change the wording for the `$1` argument and the other affected blocks (the sections referenced around lines 77-82 and 124-128) to reflect this behavior and, if implementation currently deletes the `GatewayClass` when a namespace is given, modify the delete logic so deletion of `GatewayClass` occurs only for cluster-level invocations (e.g., explicit `--cluster` or when no namespace is provided).commands/ci/ask-sippy.toml-47-59 (1)
47-59:⚠️ Potential issue | 🟠 MajorFix the quoted heredoc to enable variable expansion.
The
<<'EOF'syntax prevents shell expansion, sending the literal string$1to the Sippy API instead of the user's actual question. Change to<<EOF(without quotes around the delimiter) to allow variable substitution.Current problematic code
curl_with_token.sh https://api.cr.j7t7.p1.openshiftapps.com:6443 -s -X POST "https://sippy-auth.dptools.openshift.org/api/chat" \ -H "Content-Type: application/json" \ -d `@-` <<'EOF' { "message": "$1", "chat_history": [], "show_thinking": false, "persona": "default" } EOF🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/ask-sippy.toml` around lines 47 - 59, The heredoc passed to curl_with_token.sh is quoted (<<'EOF') which prevents shell variable expansion so the payload sends the literal "$1" instead of the caller's argument; change the heredoc delimiter to an unquoted form (<<EOF) so the shell expands $1 inside the JSON payload for the POST in the curl_with_token.sh invocation.commands/git/suggest-reviewers.toml-57-88 (1)
57-88:⚠️ Potential issue | 🟠 MajorNew untracked files are excluded from all reviewer analysis paths.
On line 58,
git status --short | grep -v '^??'discards untracked files. In Cases 1–3 (lines 61–88),git diffandgit diff --cachedare then used to collect changed files, but neither command includes untracked files. Since new files have no git blame history, OWNERS-based suggestions would be the only signal—yet untracked files are never collected, so no reviewer suggestions are made for them.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/suggest-reviewers.toml` around lines 57 - 88, The script currently filters out untracked files with the pipeline using has_uncommitted (git status --short | grep -v '^??') and then collects changed files with git diff/git diff --cached, which never include new untracked files; update the collection logic (the branches handling "Case 1/2/3" where diffs are gathered) to also run a command that lists untracked files (e.g., using git ls-files --others --exclude-standard) and merge those results with the outputs of git diff and git diff --cached, deduplicating before feeding into the OWNERS/reviewer logic so new files are considered for reviewer suggestions. Ensure the combined set is used wherever the script currently references the diff-based file lists (the variables/places that gather staged, unstaged, committed, and combined files).commands/gwapi/delete.toml-24-29 (1)
24-29:⚠️ Potential issue | 🟠 MajorThe
kubectlfallback is documented but not implemented in the actual command sequence.Tool detection (Step 1) advertises checking for
kubectlwhenocis unavailable, but Steps 3–8 hard-codeocfor all resource discovery, deletion, and verification commands. On a Kubernetes-only workstation, the flow passes tool detection and then fails immediately when executing the firstoccommand. Step 2 is the only exception, offering bothoc whoamiandkubectl cluster-infoalternatives.To support true
kubectlfallback, either remove it from Step 1 or implementkubectlalternatives throughout Steps 3–8.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/gwapi/delete.toml` around lines 24 - 29, Step 1 advertises a kubectl fallback but later Steps 3–8 hard-code oc; change the command flow to detect the available CLI once (use the existing oc whoami / kubectl cluster-info checks) and set a single TOOL variable (e.g., TOOL or kube_cmd) to either "oc" or "kubectl", then replace all hard-coded occurrances of oc in Steps 3–8 with that TOOL variable so resource discovery, deletion, and verification use the detected CLI; ensure commands referenced like oc whoami and kubectl cluster-info remain as detection alternatives and that subsequent commands (resource listing, delete, wait/check) use the unified TOOL symbol.commands/ci/trigger-postsubmit.toml-7-10 (1)
7-10:⚠️ Potential issue | 🟠 MajorFinish the Claude→Gemini conversion in this prompt.
The generated command is
ci:trigger-postsubmit, but Line 9 and the later examples/follow-ups still use/trigger-postsubmitand/query-job-status, while Lines 30-46 and 122 still address “Claude”. That leaves the generated extension internally inconsistent and can make Gemini suggest commands that do not exist under theci:namespace.Suggested prompt cleanup
-/trigger-postsubmit <job-name> <org> <repo> <base-ref> <base-sha> [ENV_VAR=value ...] +/ci:trigger-postsubmit <job-name> <org> <repo> <base-ref> <base-sha> [ENV_VAR=value ...] -9. **Offer Follow-up**: Optionally offer to query the job status using `/query-job-status` +9. **Offer Follow-up**: Optionally offer to query the job status using `/ci:query-job-status` -**Important for Claude**: +**Important for Gemini CLI**: -/trigger-postsubmit branch-ci-openshift-assisted-installer-release-4.12-images openshift assisted-installer release-4.12 7336f38f75f91a876313daacbfw97f25dfe21bbf +/ci:trigger-postsubmit branch-ci-openshift-assisted-installer-release-4.12-images openshift assisted-installer release-4.12 7336f38f75f91a876313daacbfw97f25dfe21bbfAlso applies to: 30-49, 116-128, 132-145
commands/code-review/pre-commit-review.toml-39-40 (1)
39-40:⚠️ Potential issue | 🟠 MajorThese skill lookups still point at the pre-conversion layout.
The PR output stores generated skills at repo-root
skills/with plugin-prefixed names, but this prompt still looks forskills/lang-<lang>/SKILL.mdandskills/profile-<name>/SKILL.mdrelative to a plugin root. In the generated Gemini extension those lookups miss, so language/profile-aware review silently falls back to generic mode.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/code-review/pre-commit-review.toml` around lines 39 - 40, The skill lookup paths are still using the old plugin-relative layout (checking for "skills/lang-<lang>/SKILL.md" and "skills/profile-<name>/SKILL.md"); update the logic that handles language and profile skill discovery to look in the repo-root "skills/" directory for plugin-prefixed filenames (e.g., "skills/<plugin>-lang-<lang>/SKILL.md" and "skills/<plugin>-profile-<name>/SKILL.md"), keep the same behavior when the files are missing (inform/warn and fall back to generic review), and ensure the --profile handling uses the new repo-root lookup; modify the code that constructs those paths (the skill lookup code referenced by the language/profile checks) to build and read these repo-root plugin-prefixed paths.commands/code-review/pr.toml-41-42 (1)
41-42:⚠️ Potential issue | 🟠 MajorThese skill lookups still point at the pre-conversion layout.
The generated extension stores skills at repo-root
skills/with plugin-prefixed names, but this command still searches forskills/lang-<lang>/SKILL.mdandskills/profile-<name>/SKILL.mdrelative to a plugin root. That means PR reviews will never load the converted language/profile skills and will degrade to the generic path.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/code-review/pr.toml` around lines 41 - 42, The PR review command's skill lookup still uses the old plugin-relative paths; update the lookup code in the PR review handler (e.g., the logic in processPRReviewOptions / findSkillFile / loadSkillContent) so it first checks repo-root prefixed skill names like skills/<plugin>-lang-<lang>/SKILL.md for language skills and skills/<plugin>-profile-<name>/SKILL.md for profile skills (then falls back to the generic path if not found), and ensure the content is read and stored the same way as before for use by sub-agents.commands/jira/create.toml-734-739 (1)
734-739:⚠️ Potential issue | 🟠 MajorReplace the token-shaped example before it trips secret scanning again.
Line 737 contains a realistic
sk_live_...placeholder that Gitleaks already flags as a generic API key. Keeping that in a generated artifact will create noisy security failures in CI and make real leaks harder to spot.Safer example text
Steps to reproduce: -1. Export API_KEY=sk_live_abc123xyz +1. Export API_KEY=YOUR_API_KEY🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/create.toml` around lines 734 - 739, The example includes a realistic token-shaped value ("Export API_KEY=sk_live_abc123xyz") which triggers secret scanners; update the example in the commands/jira/create.toml content to use a clearly non-secret placeholder (e.g., "Export API_KEY=YOUR_API_KEY" or "Export API_KEY=REPLACE_WITH_API_KEY") wherever the API_KEY example appears, and avoid any strings that match common secret prefixes like "sk_live_" so CI/Gitleaks no longer flags it.commands/jira/create.toml-49-55 (1)
49-55:⚠️ Potential issue | 🟠 MajorDon't use Epic Link field for Task→Story relationships; use a proper subtask or parent-child mechanism.
Line 54 maps
Task → Storytocustomfield_12311140(Epic Link), yet the validation rules (line 135) and argument documentation (line 538) both advertise "Epic or Story" as valid Task parents. Epic Link is semantically designed for Epic relationships only; using it for Task→Story creates a mismatch. Either remove Task→Story from the allowed parent types and restrict Tasks to Epic parents only, or implement Story-child Tasks as Jira subtasks using the standardparentfield.Also applies to: 132-137, 534-539
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/create.toml` around lines 49 - 55, The mapping for "Task → Story" incorrectly uses the Epic Link custom field (`additional_fields.customfield_12311140`) — update the create.toml so Tasks cannot use Epic Link as a Story parent: either remove the "Task → Story" row from the relationships table and update the related validation rules (the block around lines 132-137) and argument docs (around lines 534-539) to restrict Tasks to Epic parents only, or change Task→Story handling to create Story-child Tasks as Jira subtasks by using the standard `parent` field (replace the `additional_fields.customfield_12311140` mapping for the Task→Story case and adjust validation/docs to reflect `parent`-based subtasks).commands/git/commit-suggest.toml-40-50 (1)
40-50:⚠️ Potential issue | 🟠 MajorAdd clean worktree validation and explicit confirmation before history rewrite in Mode 2.
Mode 2 currently proceeds from message selection directly to
git commit --amendor squash operations without checking the working tree state or requiring final approval. If the repo has unrelated staged or unstaged changes, the rewrite can inadvertently fold them into history.Add a clean worktree check via
git status --porcelainbefore step 4. If the tree is not clean, stop and ask the user to stash or commit first. After suggestion selection, show the exact git command to be executed and require a separate explicit yes/no confirmation before mutating history.Suggested implementation
**Mode 2 (with N):** 1. Retrieve last N commits using `git log` 2. Parse commit messages to extract types, scopes, and descriptions +3. Verify the working tree is clean with `git status --porcelain`. + - If not clean, stop and ask the user to stash/commit unrelated changes first. -3. For **N=1**: Suggest improved rewrite - For **N≥2**: Merge commits intelligently by type priority (`fix > feat > perf > refactor > docs > test > chore`) -4. Generate 3 commit message suggestions (Recommended, Standard, Minimal) -5. Display formatted suggestions and prompt user for selection +4. For **N=1**: Suggest improved rewrite + For **N≥2**: Merge commits intelligently by type priority (`fix > feat > perf > refactor > docs > test > chore`) +5. Generate 3 commit message suggestions (Recommended, Standard, Minimal) +6. Display formatted suggestions and prompt user for selection - Ask: "Which suggestion would you like to use? (1/2/3 or skip)" - - Support responses: `1`, `use option 2`, `amend with option 3`, `skip` - - Execute `git commit --amend` (N=1) or squash operation (N≥2) if user requests + - Support responses: `1`, `use option 2`, `skip` + - Before any history rewrite, show the exact git command and require a separate explicit `yes/no` confirmation. + - Execute `git commit --amend` (N=1) or squash operation (N≥2) only after that confirmation🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/git/commit-suggest.toml` around lines 40 - 50, Mode 2 currently proceeds to history rewrites without validating the worktree or explicit user confirmation; add a pre-selection worktree cleanliness check by running "git status --porcelain" and if non-empty abort with a message to stash/commit changes, and after the user picks one of the three suggestions (Recommended/Standard/Minimal) display the exact git command that will be run (e.g. the exact git commit --amend or the squash sequence) and require a separate explicit yes/no confirmation before executing the rewrite; update the Mode 2 flow logic (the selection -> execution path) to perform the cleanliness check and then the two-step confirmation (selection then execute) so commits are not mutated unintentionally.commands/jira/categorize-activity-type.toml-115-125 (1)
115-125:⚠️ Potential issue | 🟠 MajorMove custom field to
additional_fieldsper MCP contract.The official MCP reference documentation specifies that custom fields must be updated via
additional_fields, notfields. Thefieldsparameter is reserved for standard Jira fields. Placingcustomfield_12320040infieldsviolates the documented contract and will likely cause the update to fail.Suggested fix
mcp__atlassian__jira_update_issue( issue_key="${1}", - fields={ + fields={}, + additional_fields={ "customfield_12320040": { # Activity Type field "value": "<SELECTED_ACTIVITY_TYPE>" } } )Note:
commands/jira/create-release-note.tomlhas the same issue and should be fixed separately.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/categorize-activity-type.toml` around lines 115 - 125, The MCP call mcp__atlassian__jira_update_issue currently places customfield_12320040 inside the fields argument which violates the MCP contract; move the customfield_12320040 key/value into the additional_fields payload instead of fields (i.e., stop writing customfield_12320040 under fields and add it under additional_fields with the same value structure), and make the same change in the other occurrence in create-release-note (ensure you keep the "value": "<SELECTED_ACTIVITY_TYPE>" shape when moving it).commands/etcd/analyze-performance.toml-82-93 (1)
82-93:⚠️ Potential issue | 🟠 MajorInconsistency in cluster analysis scope: log diagnostics should cover all etcd members.
Line 85 selects a single arbitrary etcd pod using
.items[0], while lines 100–150 correctly use the--clusterflag to report database statistics and health for all three members. However, lines 168–275 perform all log-based diagnostics (slow operations, disk warnings, leader changes, proposal latency, network issues) on only that one pod. This creates an analysis gap: performance degradation on the other two members will be completely missed. Refactor the log analysis section to iterate over all running etcd members (oc get pods -n openshift-etcd -l app=etcd --field-selector=status.phase=Running -o jsonpath='{.items[*].metadata.name}') and report findings per member before providing cluster-wide summary.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/etcd/analyze-performance.toml` around lines 82 - 93, The script currently sets ETCD_POD using jsonpath '.items[0]' (ETCD_POD variable) and then runs all log-based diagnostics against that single pod; change this to capture all running etcd pod names using jsonpath '.items[*].metadata.name' (e.g., ETCD_PODS) and refactor the log analysis section to loop over each pod name, running the existing per-pod diagnostics (slow operations, disk warnings, leader changes, proposal latency, network issues) for each member and printing per-member findings, then keep the existing cluster-wide checks (the commands that use --cluster) as a summary/aggregation step. Ensure references to ETCD_POD in the log analysis are replaced with the per-pod loop variable so no single-member-only commands remain.
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 9, 2026 12:15
1efc60a to
67c2b11
Compare
Contributor
There was a problem hiding this comment.
Actionable comments posted: 18
Note
Due to the large number of review comments, Critical, Major severity comments were prioritized as inline comments.
🟡 Minor comments (14)
commands/hcp/generate.toml-208-241 (1)
208-241:⚠️ Potential issue | 🟡 MinorNested code blocks will render incorrectly in markdown.
The example output section embeds a
```bashblock inside an outer```block. Standard markdown parsers will interpret the inner triple-backticks as closing the outer block, causing malformed rendering.Consider using indentation (4 spaces) for the outer block or escaping the inner backticks:
🔧 Proposed fix using indented code block for outer wrapper
**Example output:** -``` -## Summary -Creating a development AWS hosted cluster with basic configuration. - -## Generated Command -```bash + + ## Summary + Creating a development AWS hosted cluster with basic configuration. + + ## Generated Command + ```bash + hypershift create cluster aws \ + --name dev-cluster \ + ... + ``` + + ## Key Decisions + ...Alternatively, escape the inner backticks with a backslash or use a different fence character (e.g.,
~~~) for the inner block.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/hcp/generate.toml` around lines 208 - 241, The markdown example uses a fenced code block that contains another fenced block (a ```bash``` inside an outer ```), which breaks rendering; fix the "Example output" by replacing the outer fence with an indented code block (4-space indent) or by switching the inner or outer fence to a different delimiter (e.g., use ~~~ for the inner block or escape the inner backticks) so the nested command block renders correctly in the generated TOML content.commands/hcp/generate.toml-349-355 (1)
349-355:⚠️ Potential issue | 🟡 MinorSkill paths reference outdated directory structure.
The documentation references
plugins/hypershift/skills/which doesn't exist after conversion. Skills are located atskills/with plugin-prefixed names.Correct paths
To view skill details: ```bash -ls plugins/hypershift/skills/ -cat plugins/hypershift/skills/hcp-create-aws/SKILL.md -cat plugins/hypershift/skills/hcp-create-kubevirt/SKILL.md +ls skills/ +cat skills/hcp-hcp-create-aws/SKILL.md +cat skills/hcp-hcp-create-kubevirt/SKILL.md # ... etc for other providers ```'''🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/hcp/generate.toml` around lines 349 - 355, Update the documentation references that point to the now-removed plugins/hypershift/skills/ directory to the new skills/ layout and use the plugin-prefixed skill directories; e.g. replace occurrences like "plugins/hypershift/skills/" and "hcp-create-aws" with "skills/" and "hcp-hcp-create-aws" (and similarly "hcp-hcp-create-kubevirt" for kubevirt) so the examples read "ls skills/" and "cat skills/hcp-hcp-create-aws/SKILL.md" etc.commands/must-gather/ovn-dbs.toml-54-54 (1)
54-54:⚠️ Potential issue | 🟡 MinorReplace Claude reference with Gemini.
Per the PR objectives, Claude-specific references should be replaced with Gemini equivalents. This line still references "Claude" instead of "Gemini CLI".
,
📝 Proposed fix
-Claude will automatically locate it by searching for the script in the plugin installation directory, regardless of your current working directory. +Gemini CLI will automatically locate it by searching for the script in the plugin installation directory, regardless of your current working directory.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/must-gather/ovn-dbs.toml` at line 54, Update the human-readable sentence that currently says "Claude will automatically locate it by searching for the script in the plugin installation directory, regardless of your current working directory." to reference the Gemini product instead: replace "Claude" with "Gemini CLI" so it reads that Gemini CLI will automatically locate the script; edit the literal string in commands/must-gather/ovn-dbs.toml where that sentence appears.commands/jira/solve.toml-69-71 (1)
69-71:⚠️ Potential issue | 🟡 MinorFix Go syntax in example.
The example function signature uses invalid Go syntax. Parameter types should follow parameter names.
📝 Proposed fix
- - For example, a comment should not be generated for a simple function like func add(int a, b) int { return a + b} + - For example, a comment should not be generated for a simple function like func add(a, b int) int { return a + b }🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/solve.toml` around lines 69 - 71, The example under the guidance for godoc comments uses invalid Go parameter syntax for the function symbol func add; update that example so parameter names precede their types (e.g., list parameter names followed by their shared type or each name with its type) and keep the return type and body unchanged, so the example is valid Go syntax and compiles if copied.Makefile-45-47 (1)
45-47:⚠️ Potential issue | 🟡 MinorTarget help text points to the wrong generated artifacts.
convert_to_gemini.py --checkvalidates the root-level generated outputs, not agemini-extensions/directory. When this target fails, the current description will send contributors looking in the wrong place.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@Makefile` around lines 45 - 47, Update the Makefile target description for verify-gemini-sync to accurately reflect what convert_to_gemini.py --check validates: change the help text from mentioning "gemini-extensions/" to indicating it verifies the root-level generated outputs (or "generated artifacts at the repository root") so contributors are pointed to the correct location; reference the Makefile target name verify-gemini-sync and the script convert_to_gemini.py --check when making this wording change.commands/ci/list-step.toml-63-66 (1)
63-66:⚠️ Potential issue | 🟡 MinorExample command omits the
/ci:namespace.The synopsis defines
/ci:list-step, but the example uses/list-step. Copy-pasting this example will invoke the wrong command.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/list-step.toml` around lines 63 - 66, The example invocation is missing the required /ci: namespace so users will call the wrong command; update the example to use the correct command name (/ci:list-step) instead of /list-step in the snippet and any related examples so they match the synopsis (reference: the /ci:list-step command name used in the synopsis).commands/ci/list-step.toml-52-59 (1)
52-59:⚠️ Potential issue | 🟡 MinorClaude-specific wording leaked into the generated Gemini command.
The PR objective says Claude-specific references are rewritten, but this section is still titled
Important for Claude:. That makes the converted extension look only partially migrated.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/list-step.toml` around lines 52 - 59, The header "Important for Claude:" in commands/ci/list-step.toml leaks Claude-specific wording; search for that exact heading and any occurrences of "Claude" in the file (including the numbered checklist lines) and replace them with neutral, target-agnostic phrasing (e.g., "Important for agent:" or "Important for the analyzer:") and ensure the remaining checklist entries (lines starting with "1. REQUIRED:" etc.) do not contain any Claude-specific references so the converted Gemini command is fully migrated.commands/ci/trigger-postsubmit.toml-55-63 (1)
55-63:⚠️ Potential issue | 🟡 MinorImplementation step numbering error.
Step numbering jumps from step 1 to step 3, skipping step 2. This appears to be a copy-paste or editing error.
📝 Suggested fix
- $6-$N: environment variable overrides in KEY=VALUE format (optional) -3. **Construct JSON Payload**: Build the payload with refs structure: +2. **Construct JSON Payload**: Build the payload with refs structure:🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/trigger-postsubmit.toml` around lines 55 - 63, The markdown numbered steps in the comment block are misnumbered (it shows "1. **Parse Arguments**" then jumps to "3. **Construct JSON Payload**"); update the sequence so the intermediate step is "2." (or renumber the entire list) to restore proper ordering and consistency—look for the strings "1. **Parse Arguments**" and "3. **Construct JSON Payload**" in the TOML/comment and change the numbering to "2." (or sequential numbers) so the steps read 1, 2, 3 correctly.commands/ci/trigger-postsubmit.toml-7-10 (1)
7-10:⚠️ Potential issue | 🟡 MinorSynopsis command name mismatch.
The synopsis on line 9 shows
/trigger-postsubmitbut the command name defined on line 5 isci:trigger-postsubmit. This inconsistency could confuse users.📝 Suggested fix
## Synopsis-/trigger-postsubmit [ENV_VAR=value ...]
+/ci:trigger-postsubmit [ENV_VAR=value ...]🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/ci/trigger-postsubmit.toml` around lines 7 - 10, The synopsis uses the wrong command name: update the synopsis line that currently shows "/trigger-postsubmit <job-name> <org> <repo> <base-ref> <base-sha> [ENV_VAR=value ...]" to use the actual command name "ci:trigger-postsubmit <job-name> <org> <repo> <base-ref> <base-sha> [ENV_VAR=value ...]" so it matches the defined command (ci:trigger-postsubmit) and avoids user confusion.commands/must-gather/analyze.toml-46-59 (1)
46-59:⚠️ Potential issue | 🟡 MinorMissing scripts in Prerequisites documentation.
The Prerequisites section lists available scripts (lines 50-59), but the Implementation section references additional scripts that aren't documented:
analyze_prometheus.py(referenced at lines 122, 136)analyze_windows_logs.py(referenced at lines 123, 140)📝 Suggested fix
├── analyze_etcd.py -└── analyze_pvs.py +├── analyze_pvs.py +├── analyze_prometheus.py +└── analyze_windows_logs.py🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/must-gather/analyze.toml` around lines 46 - 59, The Prerequisites list in analyze.toml is missing two scripts that are actually referenced elsewhere: add analyze_prometheus.py and analyze_windows_logs.py to the documented scripts under the Analysis scripts bundle so the Implementation section references match the Prerequisites; update the list that currently enumerates analyze_clusterversion.py, analyze_clusteroperators.py, analyze_nodes.py, analyze_pods.py, analyze_network.py, analyze_ovn_dbs.py, analyze_events.py, analyze_etcd.py, analyze_pvs.py to include analyze_prometheus.py and analyze_windows_logs.py (these are the files referenced in the Implementation at lines that call/mention analyze_prometheus.py and analyze_windows_logs.py).commands/jira/clone-from-github.toml-264-266 (1)
264-266:⚠️ Potential issue | 🟡 MinorJira PAT creation URL may be incorrect for Red Hat Jira.
Line 265 references the Atlassian Cloud URL (
https://id.atlassian.com/manage-profile/security/api-tokens), but Red Hat uses on-premise Jira Server atissues.redhat.com, which has a different PAT creation flow. The note on line 266 acknowledges this but could be clearer.Consider updating to provide clearer guidance:
📝 Suggested clarification
- - Create at: https://id.atlassian.com/manage-profile/security/api-tokens (for Jira Cloud) - - Or follow your organization's Jira PAT creation process + - For Red Hat Jira: https://issues.redhat.com (Profile → Personal Access Tokens) + - For Jira Cloud: https://id.atlassian.com/manage-profile/security/api-tokens🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/clone-from-github.toml` around lines 264 - 266, Update the Jira PAT guidance under the "**Jira**: Personal Access Token" section to avoid pointing exclusively to Atlassian Cloud; replace or augment the Atlassian Cloud URL with a clear note that on-prem Red Hat Jira (issues.redhat.com) uses a different PAT/token creation flow and that users should follow their Red Hat/Jira Server instance's local security or admin docs (or contact their Jira admin) to create an access token; ensure the text mentions both the Atlassian Cloud URL (https://id.atlassian.com/manage-profile/security/api-tokens) as applicable for cloud users and a distinct sentence for Red Hat on-prem users directing them to issues.redhat.com local procedures or admin support so the guidance is unambiguous.commands/jira/grooming.toml-127-131 (1)
127-131:⚠️ Potential issue | 🟡 MinorUnclosed code block in example 12.
The code block for example 12 is missing its closing triple backticks, which will cause Markdown rendering issues.
📝 Proposed fix
12. **Combine Label and status filters**: ```bash /jira:grooming OCPBUGS last-week --label "performance" --status "NEW" + ``` ## Output Format🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@commands/jira/grooming.toml` around lines 127 - 131, The Markdown example labeled "Combine Label and status filters" contains an unclosed code block for the bash snippet "/jira:grooming OCPBUGS last-week --label \"performance\" --status \"NEW\""; fix it by adding the missing closing triple backticks (```) immediately after that command so the fenced code block is properly terminated and renders correctly in the grooming.toml example.commands/ci/analyze-prow-job-test-failure.toml-26-41 (1)
26-41:⚠️ Potential issue | 🟡 MinorInconsistent command names in usage examples.
The examples reference
/ci:analyze-test-failure(Lines 28, 36), but the actual command name defined at Line 5 is/ci:analyze-prow-job-test-failure. This inconsistency will confuse users.📝 Proposed fix
**Default (comprehensive analysis)**: ```text -/ci:analyze-test-failure <url> <test-name> +/ci:analyze-prow-job-test-failure <url> <test-name>
- Detects must-gather availability
- Prompts user whether to include cluster diagnostics
- Provides correlated test + cluster analysis
Fast mode (skip must-gather):
-/ci:analyze-test-failure <url> <test-name> --fast +/ci:analyze-prow-job-test-failure <url> <test-name> --fast</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@commands/ci/analyze-prow-job-test-failure.tomlaround lines 26 - 41, The
usage examples incorrectly show the command as /ci:analyze-test-failure; update
both examples to use the actual command name /ci:analyze-prow-job-test-failure
(including the fast-mode example with --fast) so the displayed invocation
matches the implemented command and avoids confusion when users copy/paste.</details> </blockquote></details> <details> <summary>commands/etcd/analyze-performance.toml-29-55 (1)</summary><blockquote> `29-55`: _⚠️ Potential issue_ | _🟡 Minor_ **Add `jq` and `bc` to the prerequisites section.** The documented prerequisites only mention OpenShift CLI (`oc`), but the implementation requires `jq` (lines 111–139 for JSON parsing) and `bc` (lines 293–338 for floating-point arithmetic). On a clean admin host, the script will fail with `command not found` errors after passing initial checks. Update the prerequisites and regenerate the plugin spec. <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@commands/etcd/analyze-performance.tomlaround lines 29 - 55, Update the "##
Prerequisites" section to list jq and bc as required tools (alongside oc),
because the analyze-performance implementation uses jq for JSON parsing and bc
for floating-point arithmetic; add short install/verify hints for both (e.g.,
package names and a verification command likejq --versionandbc --version)
and then regenerate the plugin spec so the updated TOML is published.</details> </blockquote></details> </blockquote></details> <details> <summary>🧹 Nitpick comments (3)</summary><blockquote> <details> <summary>README.md (1)</summary><blockquote> `71-74`: **Add language specifier to fenced code block.** For consistency with other code blocks in this file and to satisfy markdownlint (MD040), specify a language for this code block. <details> <summary>📝 Proposed fix</summary> ```diff **Use the commands:** -``` +```bash /jira:solve OCPBUGS-12345 origin</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@README.mdaround lines 71 - 74, The fenced code block containing the command
"/jira:solve OCPBUGS-12345 origin" in README.md lacks a language specifier;
update that fenced block to include a language token (e.g., "bash") so it reads
likebash ...to match other blocks and satisfy markdownlint MD040,
ensuring the block around the string "/jira:solve OCPBUGS-12345 origin" is
changed accordingly.</details> </blockquote></details> <details> <summary>commands/ci/trigger-periodic.toml (1)</summary><blockquote> `95-101`: **Clarify AI assistant references.** The "Important for Claude" section header at line 95 seems inconsistent with this being a Gemini extension. Line 97 correctly references "Gemini CLI" but the section title and line 96's "Skill tool" reference may cause confusion about which AI assistant is being addressed. Consider updating the section header and clarifying the context if this guidance applies to Gemini CLI users. <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@commands/ci/trigger-periodic.toml` around lines 95 - 101, Update the section header and wording that currently reads "Important for Claude" to explicitly target Gemini CLI users (e.g., "Important for Gemini") and reword any mentions of "Skill tool" to the correct invocation method for the Gemini environment; ensure the instructions reference the required skill name "ci:oc-auth", the script "curl_with_token.sh", and the follow-up action "/query-job-status" so it's clear the steps are for Gemini CLI (load ci:oc-auth, locate curl_with_token.sh, parse JSON for execution ID, display it, and offer to check job status). ``` </details> </blockquote></details> <details> <summary>commands/jira/update-weekly-status.toml (1)</summary><blockquote> `276-283`: **Hardcoded custom field ID may require documentation.** The custom field ID `customfield_12320841` is hardcoded for the Status Summary field. While this is acceptable for organization-specific tooling, consider adding a comment in the prompt noting that this field ID is specific to the Red Hat Jira instance and may need adjustment for other environments. <details> <summary>💡 Suggested improvement</summary> ```diff Use `mcp__atlassian-mcp__jira_update_issue`: +**Note**: The custom field ID below (`customfield_12320841`) is specific to the Red Hat Jira instance for the "Status Summary" field. Other Jira instances may use different field IDs. + ```json { "issue_key": "{ISSUE-KEY}", ``` </details> <details> <summary>🤖 Prompt for AI Agents</summary> ``` Verify each finding against the current code and only fix it if needed. In `@commands/jira/update-weekly-status.toml` around lines 276 - 283, The JSON payload hardcodes the Jira custom field ID customfield_12320841 for the Status Summary; update the prompt or nearby documentation in update-weekly-status.toml to add a clear comment that customfield_12320841 is Red Hat–specific and may need to be changed for other Jira instances, and optionally indicate where to configure or override this field ID (e.g., via environment variable or template parameter) so maintainers know how to adapt it. ``` </details> </blockquote></details> </blockquote></details> <details> <summary>🤖 Prompt for all review comments with AI agents</summary>Verify each finding against the current code and only fix it if needed.
Inline comments:
In@commands/bigquery/analyze-usage.toml:
- Around line 64-68: The "Offer to Save Report" step that suggests saving to
bigquery-usage--.md and calls the Write tool must not
save sensitive identifiers or query previews without explicit user consent and
redaction; update the save flow to (1) add a redaction mode that strips or masks
user/service-account identifiers and query snippets before writing, (2) present
a clear explicit warning prompt summarizing what sensitive fields will be saved
and require an affirmative confirmation (e.g., "I understand the risks and want
to save unredacted"), and (3) default to saving a redacted report unless the
user explicitly chooses unredacted and confirms; make these changes in the code
path that builds the suggested filename and invokes the Write tool so the
prompt, redaction logic, and confirmed write are enforced before any file is
created.In
@commands/ci/add-debug-wait.toml:
- Around line 443-455: The script currently hard-codes pushing to "origin" and
the compare URL to
"openshift/release@master...${branch_name}", which
breaks fork workflows; change the flow to determine or prompt for the push
remote (e.g., a variable like push_remote) and derive the repository owner/name
from that remote (or ask the user if ambiguous) and then use that push_remote
and derived repo owner/name to construct both the git push command and the
compare URL instead of using "origin" and "openshift/release"; update
occurrences of git push origin "${branch_name}" and the printed compare URL to
use the new push_remote and repo variables.In
@commands/ci/payload-agent.toml:
- Around line 20-25: Summary: The HIGH-confidence action path is not idempotent
and can create duplicate TRT JIRA bugs, revert PRs, and validation-job triggers
when re-run. Fix: implement the same persistence/guard pattern used by the
MEDIUM-confidence flow—before creating a TRT JIRA bug, opening a revert PR, or
triggering payload validation jobs, check for an existing marker/tracking
artifact (the "tracking file" concept used by Phase 1/Phase 2) tied to the
payload identifier; if the marker exists, skip creation and instead reconcile or
report status. Locate the HIGH-confidence branch that performs "Automatically
creates a TRT JIRA bug, opens a revert PR, and triggers payload validation jobs"
and add existence checks and consistent marker writes equivalent to the
medium-confidence tracking-file logic so repeated runs are no-ops.In
@commands/ci/trigger-presubmit.toml:
- Around line 8-10: Update all user-visible command references to include the
namespace prefix so they match the actual command names: replace occurrences of
"/trigger-presubmit" with "/ci:trigger-presubmit" and "/query-job-status" with
"/ci:query-job-status" (and similarly for any other command mentions) in the
synopsis, examples, and follow-up steps; ensure the displayed command strings
match the defined command symbol ci:trigger-presubmit and ci:query-job-status so
generated docs/examples are accurate.In
@commands/code-review/pr.toml:
- Around line 41-42: The lookup still only checks plugin-root paths
skills/lang-<lang>/SKILL.mdandskills/profile-<name>/SKILL.md, so
--languageand--profilemiss converted, repo-level skills; update the
resolver (the code that handles the--languageand--profileflags) to first
attempt the original plugin-root paths and then fall back to repo-level
skills/directory using the plugin-prefixed filenames the converter produces
(e.g.skills/<plugin>-lang-<lang>/SKILL.mdand
skills/<plugin>-profile-<name>/SKILL.md), reading and storing the content if
found and warning/continuing only if both checks fail.In
@commands/compliance/analyze-cve.toml:
- Line 48: The generated TOML contains relative skill links like the string
"cve-intelligence-gathering"
that point to commands/skills/... instead of the top-level skills tree; update
the generator that emits commands/compliance/analyze-cve.toml (and other
affected sections producing lines similar to the "Skill:
") to rewrite those hrefs to
"skills//SKILL.md" (i.e., remove the leading "../") so generated
artifacts reference the top-level skills directory consistently across all
occurrences (also fix the same pattern at the other noted locations, e.g., lines
around 62-63 and 98).In
@commands/etcd/analyze-performance.toml:
- Around line 82-94: The script currently selects a single pod via ETCD_POD and
later pulls logs only from that pod; change this to collect logs from all etcd
members by replacing the single ETCD_POD selection with ETCD_PODS="$(oc get pods
-n openshift-etcd -l app=etcd --field-selector=status.phase=Running -o
jsonpath='{.items[*].metadata.name}')" and loop over each pod name to run oc
logs (the commands that reference ETCD_POD and the oc logs invocation)
aggregating per-member logs; also update the source plugin prompt text (the
artifact generation input) to state explicitly that logs must be gathered and
analyzed across all etcd members (cluster-wide) so sections that summarize
cluster state use the aggregated logs rather than a single pod’s logs.In
@commands/etcd/health-check.toml:
- Around line 152-160: Initialize a WARNINGS counter early (e.g., set
WARNINGS=0) and increment it each time a warning branch is triggered: after the
quorum check that uses MEMBER_COUNT, after the unstarted-members check that uses
UNSTARTED and MEMBER_LIST, and similarly in the other warning branches (the ones
around lines referencing dns/port or health checks). Concretely, add
WARNINGS=$((WARNINGS + 1)) immediately after each echo "WARNING: ..." so the
final summary that tests $WARNINGS works reliably and avoids "integer expression
expected".- Around line 46-52: Add explicit bash argument parsing and initialization for
the missing flags and counters: initialize VERBOSE=0 and WARNINGS=0 at the top
of the health-check script in plugins/etcd/commands/health-check.md, add a
command-line parsing loop that recognizes --verbose (sets VERBOSE=1) and any
other flags that should affect WARNINGS, and ensure any warning-producing checks
increment WARNINGS when they detect issues (so the summary's $WARNINGS reflects
real counts). After updating the script, regenerate the TOML so the documented
--verbose argument and summary counters match the implementation.In
@commands/git/suggest-reviewers.toml:
- Around line 95-120: Replace Claude-specific environment variable references
${CLAUDE_PLUGIN_ROOT}in the suggested commands with a Gemini-compatible or
neutral variable (e.g.,${GEMINI_PLUGIN_ROOT}or${PLUGIN_ROOT}) so the
helper script path
${CLAUDE_PLUGIN_ROOT}/skills/suggest-reviewers/analyze_blame.pybecomes
${GEMINI_PLUGIN_ROOT}/skills/suggest-reviewers/analyze_blame.py(or
${PLUGIN_ROOT}/skills/suggest-reviewers/analyze_blame.py) everywhere it
appears; update all occurrences in the file (the three command examples that
reference the helper script) to use the chosen variable consistently and ensure
the displayed bash examples use the same replacement.In
@commands/jira/backlog.toml:
- Around line 44-52: The content mixes Gemini and Claude setup flows—keep one
consistent (prefer Gemini) by replacing all Claude-specific instructions: change
any "claude mcp add" to "gemini mcp add", update the config path
"/.config/claude-code/mcp.json" to the Gemini equivalent (e.g./.config/gemini/mcp.json"), and ensure the earlier instruction to restart the
"
Gemini CLI is followed by the Gemini-specific command and config references;
apply the same normalization for the other affected block mentioned (the later
115-145 range).- Around line 505-512: Replace the unsafe inline PAT in the curl example with an
env-based or wrapper pattern: instruct users to export JIRA_PERSONAL_TOKEN (or
use the existing JIRA_PERSONAL_TOKEN/JIRA_URL variables) and run curl with the
header built from the environment (e.g., using -H "Authorization: Bearer
$JIRA_PERSONAL_TOKEN" or recommend the provided CLI wrapper) instead of
embedding the token on the command line; update the troubleshooting bullet that
currently shows curl -H "Authorization: Bearer YOUR_TOKEN"
YOUR_JIRA_URL/rest/api/2/myself to use JIRA_PERSONAL_TOKEN and JIRA_URL
environment variables and a wrapper/env-based example to avoid shell-history
token leakage.In
@commands/jira/generate-test-plan.toml:
- Around line 17-18: The command in generate-test-plan.toml hard-codes the Jira
host in the curl REST call ("https://issues.redhat.com/rest/api/2/issue/{$1}")
and in the WebFetch scraping step, so it won't respect the shared Jira host
configuration; update both places to use the shared Jira host configuration
(e.g., a common variable/env like JIRA_HOST or the project's jiraHost config)
instead of the literal "issues.redhat.com", and ensure the WebFetch URL and the
REST API URL are constructed from that single config value so the command works
against other Jira instances.In
@commands/jira/issues-by-component.toml:
- Around line 124-128: The JQL builder currently injects user tokens like
last-week,last-2-weeks, andlast-month(from the parser reading
$2/time-period) directly into the templatecreated >= -{time-period},
producing invalid JQL; update the code that prepares thetime-periodvariable
(the parser/templating step that reads$2) to map those friendly tokens to
valid JQL values (e.g.,last-week->7dor compute explicit ISO dates like
YYYY/MM/DDranges) and accept customYYYY-MM-DD:YYYY-MM-DDby converting it
into an explicitcreated >= "YYYY-MM-DD" AND created <= "YYYY-MM-DD"clause
(orcreated >= -Ndfor relative ranges). Apply the same conversion logic where
the template injection is performed so the final JQL uses valid duration syntax
or explicit dates instead of-last-week.In
@commands/jira/setup-gh2jira.toml:
- Around line 169-173: The curl examples expose secrets via command-line
arguments; instead read the tokens from the environment and supply the
Authorization header via stdin or a temporary file so the token never appears in
the process argument list—replace the two direct-argument uses (the curl call
that uses GITHUB_TOKEN and the curl call that uses JIRA_TOKEN) with a pattern
that builds the header from the environment and feeds it to curl via stdin or a
secure temp file (e.g., echo/printf the header from $GITHUB_TOKEN or $JIRA_TOKEN
and pipe it to curl using curl's "@-" or file-header mechanism), ensuring the
shell never places the raw token text into the command-line arguments.In
@commands/lvms/analyze.toml:
- Around line 159-172: The built-in file reads in commands/lvms/analyze.toml
currently hard-code the namespace "openshift-lvm-storage" for files like
"{must-gather-path}/namespaces/openshift-lvm-storage/pods.yaml" and
"{must-gather-path}/namespaces/openshift-lvm-storage/events.yaml", which drops
support for older "openshift-storage" must-gathers; update the logic/text so the
analyzer tries both namespaces (e.g., attempt
"{must-gather-path}/namespaces/openshift-lvm-storage/..." and if missing fall
back to "{must-gather-path}/namespaces/openshift-storage/..."), or iterate over
a namespaces list before giving up, and ensure the LVMCluster/LVMVolumeGroup
find commands likewise check both names when the optional Python script is
absent.- Around line 550-558: Step 4 uses the same namespace-wide delete command as
Step 2 and therefore restarts every vg-manager pod; replace that line so it
targets only the vg-manager pod running on worker-0. Change the command string
"oc delete pod -n openshift-lvm-storage -l
app.kubernetes.io/component=vg-manager" in Step 4 to a node-scoped delete that
selects the pod on worker-0 (for example by adding a field selector like
--field-selector spec.nodeName=worker-0 or by resolving the pod name on worker-0
and deleting that single pod) so only the vg-manager on worker-0 is restarted.In
@commands/must-gather/windows.toml:
- Around line 7-10: The documented flags (--component, --errors-only,
--max-errors) are not passed to the implementation and the synopsis omits
--max-errors; update the CLI invocation in commands/must-gather/windows.toml so
the command calls analyze_windows_logs.py with the flags forwarded (e.g.,
include --component, --errors-only, --max-errors when present) and update the
synopsis block to list --max-errors as well; search for any other occurrences
mentioned (around lines referenced) and ensure all places that launch
analyze_windows_logs.py or document the /must-gather:windows command
consistently accept and forward those three flags.
Minor comments:
In@commands/ci/analyze-prow-job-test-failure.toml:
- Around line 26-41: The usage examples incorrectly show the command as
/ci:analyze-test-failure; update both examples to use the actual command name
/ci:analyze-prow-job-test-failure (including the fast-mode example with --fast)
so the displayed invocation matches the implemented command and avoids confusion
when users copy/paste.In
@commands/ci/list-step.toml:
- Around line 63-66: The example invocation is missing the required /ci:
namespace so users will call the wrong command; update the example to use the
correct command name (/ci:list-step) instead of /list-step in the snippet and
any related examples so they match the synopsis (reference: the /ci:list-step
command name used in the synopsis).- Around line 52-59: The header "Important for Claude:" in
commands/ci/list-step.toml leaks Claude-specific wording; search for that exact
heading and any occurrences of "Claude" in the file (including the numbered
checklist lines) and replace them with neutral, target-agnostic phrasing (e.g.,
"Important for agent:" or "Important for the analyzer:") and ensure the
remaining checklist entries (lines starting with "1. REQUIRED:" etc.) do not
contain any Claude-specific references so the converted Gemini command is fully
migrated.In
@commands/ci/trigger-postsubmit.toml:
- Around line 55-63: The markdown numbered steps in the comment block are
misnumbered (it shows "1. Parse Arguments" then jumps to "3. Construct
JSON Payload"); update the sequence so the intermediate step is "2." (or
renumber the entire list) to restore proper ordering and consistency—look for
the strings "1. Parse Arguments" and "3. Construct JSON Payload" in the
TOML/comment and change the numbering to "2." (or sequential numbers) so the
steps read 1, 2, 3 correctly.- Around line 7-10: The synopsis uses the wrong command name: update the
synopsis line that currently shows "/trigger-postsubmit
[ENV_VAR=value ...]" to use the actual command name
"ci:trigger-postsubmit
[ENV_VAR=value ...]" so it matches the defined command (ci:trigger-postsubmit)
and avoids user confusion.In
@commands/etcd/analyze-performance.toml:
- Around line 29-55: Update the "## Prerequisites" section to list jq and bc as
required tools (alongside oc), because the analyze-performance implementation
uses jq for JSON parsing and bc for floating-point arithmetic; add short
install/verify hints for both (e.g., package names and a verification command
likejq --versionandbc --version) and then regenerate the plugin spec so
the updated TOML is published.In
@commands/hcp/generate.toml:
- Around line 208-241: The markdown example uses a fenced code block that
contains another fenced block (abashinside an outer ```), which breaks
rendering; fix the "Example output" by replacing the outer fence with an
indented code block (4-space indent) or by switching the inner or outer fence to
a different delimiter (e.g., use ~~~ for the inner block or escape the inner
backticks) so the nested command block renders correctly in the generated TOML
content.- Around line 349-355: Update the documentation references that point to the
now-removed plugins/hypershift/skills/ directory to the new skills/ layout and
use the plugin-prefixed skill directories; e.g. replace occurrences like
"plugins/hypershift/skills/" and "hcp-create-aws" with "skills/" and
"hcp-hcp-create-aws" (and similarly "hcp-hcp-create-kubevirt" for kubevirt) so
the examples read "ls skills/" and "cat skills/hcp-hcp-create-aws/SKILL.md" etc.In
@commands/jira/clone-from-github.toml:
- Around line 264-266: Update the Jira PAT guidance under the "Jira:
Personal Access Token" section to avoid pointing exclusively to Atlassian Cloud;
replace or augment the Atlassian Cloud URL with a clear note that on-prem Red
Hat Jira (issues.redhat.com) uses a different PAT/token creation flow and that
users should follow their Red Hat/Jira Server instance's local security or admin
docs (or contact their Jira admin) to create an access token; ensure the text
mentions both the Atlassian Cloud URL
(https://id.atlassian.com/manage-profile/security/api-tokens) as applicable for
cloud users and a distinct sentence for Red Hat on-prem users directing them to
issues.redhat.com local procedures or admin support so the guidance is
unambiguous.In
@commands/jira/grooming.toml:
- Around line 127-131: The Markdown example labeled "Combine Label and status
filters" contains an unclosed code block for the bash snippet "/jira:grooming
OCPBUGS last-week --label "performance" --status "NEW""; fix it by adding
the missing closing triple backticks (```) immediately after that command so the
fenced code block is properly terminated and renders correctly in the
grooming.toml example.In
@commands/jira/solve.toml:
- Around line 69-71: The example under the guidance for godoc comments uses
invalid Go parameter syntax for the function symbol func add; update that
example so parameter names precede their types (e.g., list parameter names
followed by their shared type or each name with its type) and keep the return
type and body unchanged, so the example is valid Go syntax and compiles if
copied.In
@commands/must-gather/analyze.toml:
- Around line 46-59: The Prerequisites list in analyze.toml is missing two
scripts that are actually referenced elsewhere: add analyze_prometheus.py and
analyze_windows_logs.py to the documented scripts under the Analysis scripts
bundle so the Implementation section references match the Prerequisites; update
the list that currently enumerates analyze_clusterversion.py,
analyze_clusteroperators.py, analyze_nodes.py, analyze_pods.py,
analyze_network.py, analyze_ovn_dbs.py, analyze_events.py, analyze_etcd.py,
analyze_pvs.py to include analyze_prometheus.py and analyze_windows_logs.py
(these are the files referenced in the Implementation at lines that call/mention
analyze_prometheus.py and analyze_windows_logs.py).In
@commands/must-gather/ovn-dbs.toml:
- Line 54: Update the human-readable sentence that currently says "Claude will
automatically locate it by searching for the script in the plugin installation
directory, regardless of your current working directory." to reference the
Gemini product instead: replace "Claude" with "Gemini CLI" so it reads that
Gemini CLI will automatically locate the script; edit the literal string in
commands/must-gather/ovn-dbs.toml where that sentence appears.In
@Makefile:
- Around line 45-47: Update the Makefile target description for
verify-gemini-sync to accurately reflect what convert_to_gemini.py --check
validates: change the help text from mentioning "gemini-extensions/" to
indicating it verifies the root-level generated outputs (or "generated artifacts
at the repository root") so contributors are pointed to the correct location;
reference the Makefile target name verify-gemini-sync and the script
convert_to_gemini.py --check when making this wording change.
Nitpick comments:
In@commands/ci/trigger-periodic.toml:
- Around line 95-101: Update the section header and wording that currently reads
"Important for Claude" to explicitly target Gemini CLI users (e.g., "Important
for Gemini") and reword any mentions of "Skill tool" to the correct invocation
method for the Gemini environment; ensure the instructions reference the
required skill name "ci:oc-auth", the script "curl_with_token.sh", and the
follow-up action "/query-job-status" so it's clear the steps are for Gemini CLI
(load ci:oc-auth, locate curl_with_token.sh, parse JSON for execution ID,
display it, and offer to check job status).In
@commands/jira/update-weekly-status.toml:
- Around line 276-283: The JSON payload hardcodes the Jira custom field ID
customfield_12320841 for the Status Summary; update the prompt or nearby
documentation in update-weekly-status.toml to add a clear comment that
customfield_12320841 is Red Hat–specific and may need to be changed for other
Jira instances, and optionally indicate where to configure or override this
field ID (e.g., via environment variable or template parameter) so maintainers
know how to adapt it.In
@README.md:
- Around line 71-74: The fenced code block containing the command "/jira:solve
OCPBUGS-12345 origin" in README.md lacks a language specifier; update that
fenced block to include a language token (e.g., "bash") so it reads likebash ...to match other blocks and satisfy markdownlint MD040, ensuring the block
around the string "/jira:solve OCPBUGS-12345 origin" is changed accordingly.</details> --- <details> <summary>ℹ️ Review info</summary> <details> <summary>⚙️ Run configuration</summary> **Configuration used**: Path: .coderabbit.yaml **Review profile**: CHILL **Plan**: Pro **Run ID**: `477f4806-0012-4a86-bb20-772e858670e3` </details> <details> <summary>📥 Commits</summary> Reviewing files that changed from the base of the PR and between 1efc60a88fc70793c0d48ba2c0227e04db8ae764 and 67c2b11fafa9aa57e90321104aadc927e1dd2277. </details> <details> <summary>📒 Files selected for processing (300)</summary> * `.github/workflows/verify-gemini-sync.yml` * `GEMINI.md` * `Makefile` * `README.md` * `commands/agendas/outcome-refinement.toml` * `commands/bigquery/analyze-usage.toml` * `commands/ci/add-debug-wait.toml` * `commands/ci/analyze-payload.toml` * `commands/ci/analyze-pr-reverts.toml` * `commands/ci/analyze-prow-job-install-failure.toml` * `commands/ci/analyze-prow-job-resource.toml` * `commands/ci/analyze-prow-job-test-failure.toml` * `commands/ci/analyze-regression.toml` * `commands/ci/ask-sippy.toml` * `commands/ci/check-if-jira-regression-is-ongoing.toml` * `commands/ci/extract-prow-job-must-gather.toml` * `commands/ci/fetch-payloads.toml` * `commands/ci/fetch-test-report.toml` * `commands/ci/list-step.toml` * `commands/ci/list-unstable-tests.toml` * `commands/ci/payload-agent.toml` * `commands/ci/query-job-status.toml` * `commands/ci/query-test-result.toml` * `commands/ci/revert-pr.toml` * `commands/ci/trigger-periodic.toml` * `commands/ci/trigger-postsubmit.toml` * `commands/ci/trigger-presubmit.toml` * `commands/code-review/pr.toml` * `commands/code-review/pre-commit-review.toml` * `commands/compliance/analyze-cve.toml` * `commands/container-image/compare.toml` * `commands/container-image/inspect.toml` * `commands/container-image/tags.toml` * `commands/doc/note.toml` * `commands/etcd/analyze-performance.toml` * `commands/etcd/health-check.toml` * `commands/git/backport.toml` * `commands/git/bisect.toml` * `commands/git/branch-cleanup.toml` * `commands/git/cherry-pick-by-patch.toml` * `commands/git/commit-suggest.toml` * `commands/git/debt-scan.toml` * `commands/git/fix-cherrypick-robot-pr.toml` * `commands/git/redescribe.toml` * `commands/git/suggest-reviewers.toml` * `commands/git/summary.toml` * `commands/golang/lint-fix.toml` * `commands/gwapi/check.toml` * `commands/gwapi/delete.toml` * `commands/gwapi/install.toml` * `commands/hcp/cluster-health-check.toml` * `commands/hcp/generate.toml` * `commands/hello-world/echo.toml` * `commands/jira/backlog.toml` * `commands/jira/categorize-activity-type.toml` * `commands/jira/clone-from-github.toml` * `commands/jira/create-release-note.toml` * `commands/jira/create.toml` * `commands/jira/generate-feature-doc.toml` * `commands/jira/generate-test-plan.toml` * `commands/jira/grooming.toml` * `commands/jira/issues-by-component.toml` * `commands/jira/reconcile-github.toml` * `commands/jira/setup-gh2jira.toml` * `commands/jira/solve.toml` * `commands/jira/status-rollup.toml` * `commands/jira/update-weekly-status.toml` * `commands/jira/validate-blockers.toml` * `commands/lvms/analyze.toml` * `commands/must-gather/analyze.toml` * `commands/must-gather/ovn-dbs.toml` * `commands/must-gather/windows.toml` * `commands/node-tuning/analyze-node-tuning.toml` * `commands/node-tuning/generate-tuned-profile.toml` * `commands/node/cluster-node-health-check.toml` * `commands/olm-team/configure-agent.toml` * `commands/olm-team/dev-setup.toml` * `commands/olm-team/ep-watch.toml` * `commands/olm/approve.toml` * `commands/olm/catalog.toml` * `commands/olm/debug.toml` * `commands/olm/diagnose.toml` * `commands/olm/install.toml` * `commands/olm/list.toml` * `commands/olm/opm.toml` * `commands/olm/search.toml` * `commands/olm/status.toml` * `commands/olm/uninstall.toml` * `commands/olm/upgrade.toml` * `commands/openshift/add-enhancement.toml` * `commands/openshift/bootstrap-om.toml` * `commands/openshift/bump-deps.toml` * `commands/openshift/cluster-health-check.toml` * `commands/openshift/crd-review.toml` * `commands/openshift/create-cluster.toml` * `commands/openshift/destroy-cluster.toml` * `commands/openshift/expand-test-case.toml` * `commands/openshift/ironic-status.toml` * `commands/openshift/new-e2e-test.toml` * `commands/openshift/node-kernel-conntrack.toml` * `commands/openshift/node-kernel-ip.toml` * `commands/openshift/node-kernel-iptables.toml` * `commands/openshift/node-kernel-nft.toml` * `commands/openshift/rebase.toml` * `commands/openshift/review-test-cases.toml` * `commands/openshift/visualize-ovn-topology.toml` * `commands/origin/two-node-origin-pr-helper.toml` * `commands/ote-migration/migrate.toml` * `commands/session/save-session.toml` * `commands/sosreport/analyze.toml` * `commands/sosreport/ovs-db.toml` * `commands/teams/coderabbit-adoption-report.toml` * `commands/teams/coderabbit-inheritance-scanner.toml` * `commands/teams/health-check-jiras.toml` * `commands/teams/health-check-regressions.toml` * `commands/teams/health-check.toml` * `commands/teams/list-components.toml` * `commands/teams/list-jiras.toml` * `commands/teams/list-regressions.toml` * `commands/teams/list-teams.toml` * `commands/test-coverage/analyze.toml` * `commands/test-coverage/gaps.toml` * `commands/testing/mutation-test.toml` * `commands/utils/address-reviews.toml` * `commands/utils/auto-approve-konflux-prs.toml` * `commands/utils/generate-test-plan.toml` * `commands/utils/gh-attention.toml` * `commands/utils/placeholder.toml` * `commands/utils/process-renovate-pr.toml` * `commands/utils/review-ai-helpers-overlap.toml` * `commands/utils/review-security.toml` * `commands/workspaces/create.toml` * `commands/workspaces/delete.toml` * `commands/yaml/docs.toml` * `gemini-extension.json` * `scripts/convert_to_gemini.py` * `skills/bigquery-analyze-usage/SKILL.md` * `skills/ci-analyze-payload/SKILL.md` * `skills/ci-bisect-payload-suspects/SKILL.md` * `skills/ci-fetch-jira-issue/README.md` * `skills/ci-fetch-jira-issue/SKILL.md` * `skills/ci-fetch-jira-issue/fetch_jira_issue.py` * `skills/ci-fetch-new-prs-in-payload/SKILL.md` * `skills/ci-fetch-new-prs-in-payload/fetch_new_prs_in_payload.py` * `skills/ci-fetch-payloads/SKILL.md` * `skills/ci-fetch-payloads/fetch_payloads.py` * `skills/ci-fetch-prowjob-json/SKILL.md` * `skills/ci-fetch-regression-details/README.md` * `skills/ci-fetch-regression-details/SKILL.md` * `skills/ci-fetch-regression-details/fetch_regression_details.py` * `skills/ci-fetch-related-triages/SKILL.md` * `skills/ci-fetch-related-triages/fetch_related_triages.py` * `skills/ci-fetch-releases/SKILL.md` * `skills/ci-fetch-releases/fetch_releases.py` * `skills/ci-fetch-test-report/SKILL.md` * `skills/ci-fetch-test-report/fetch_test_report.py` * `skills/ci-fetch-test-runs/README.md` * `skills/ci-fetch-test-runs/SKILL.md` * `skills/ci-fetch-test-runs/fetch_test_runs.py` * `skills/ci-oc-auth/README.md` * `skills/ci-oc-auth/SKILL.md` * `skills/ci-oc-auth/curl_with_token.sh` * `skills/ci-payload-agent/SKILL.md` * `skills/ci-prow-job-analyze-install-failure/SKILL.md` * `skills/ci-prow-job-analyze-metal-install-failure/SKILL.md` * `skills/ci-prow-job-analyze-resource/CHANGELOG.md` * `skills/ci-prow-job-analyze-resource/README.md` * `skills/ci-prow-job-analyze-resource/SCRIPTS.md` * `skills/ci-prow-job-analyze-resource/SKILL.md` * `skills/ci-prow-job-analyze-resource/create_context_html_files.py` * `skills/ci-prow-job-analyze-resource/create_inline_html_files.py` * `skills/ci-prow-job-analyze-resource/generate_html_report.py` * `skills/ci-prow-job-analyze-resource/generate_report.py` * `skills/ci-prow-job-analyze-resource/parse_all_logs.py` * `skills/ci-prow-job-analyze-resource/parse_audit_logs.py` * `skills/ci-prow-job-analyze-resource/parse_pod_logs.py` * `skills/ci-prow-job-analyze-resource/parse_url.py` * `skills/ci-prow-job-analyze-resource/prow_job_resource_grep.sh` * `skills/ci-prow-job-analyze-resource/report_template.html` * `skills/ci-prow-job-analyze-test-failure/README.md` * `skills/ci-prow-job-analyze-test-failure/SKILL.md` * `skills/ci-prow-job-artifact-search/SKILL.md` * `skills/ci-prow-job-artifact-search/prow_job_artifact_search.py` * `skills/ci-prow-job-extract-must-gather/CHANGELOG.md` * `skills/ci-prow-job-extract-must-gather/README.md` * `skills/ci-prow-job-extract-must-gather/SKILL.md` * `skills/ci-prow-job-extract-must-gather/extract_archives.py` * `skills/ci-prow-job-extract-must-gather/generate_html_report.py` * `skills/ci-revert-pr/SKILL.md` * `skills/ci-set-release-blocker/SKILL.md` * `skills/ci-set-release-blocker/set_release_blocker.py` * `skills/ci-stage-payload-reverts/SKILL.md` * `skills/ci-triage-regression/README.md` * `skills/ci-triage-regression/SKILL.md` * `skills/ci-triage-regression/triage_regression.py` * `skills/ci-trigger-payload-job/SKILL.md` * `skills/code-review-lang-go/SKILL.md` * `skills/code-review-profile-hypershift/SKILL.md` * `skills/compliance-call-graph-analysis/SKILL.md` * `skills/compliance-codebase-impact-analysis/SKILL.md` * `skills/compliance-cve-intelligence-gathering/SKILL.md` * `skills/compliance-remediation-planning/SKILL.md` * `skills/git-suggest-reviewers/SKILL.md` * `skills/git-suggest-reviewers/analyze_blame.py` * `skills/golang-lint/SKILL.md` * `skills/hcp-hcp-create-agent/SKILL.md` * `skills/hcp-hcp-create-aws/SKILL.md` * `skills/hcp-hcp-create-azure/SKILL.md` * `skills/hcp-hcp-create-kubevirt/SKILL.md` * `skills/hcp-hcp-create-openstack/SKILL.md` * `skills/hcp-hcp-create-powervs/SKILL.md` * `skills/jira-categorize-activity-type/SKILL.md` * `skills/jira-cntrlplane/SKILL.md` * `skills/jira-create-bug/SKILL.md` * `skills/jira-create-epic/SKILL.md` * `skills/jira-create-feature-request/SKILL.md` * `skills/jira-create-feature/SKILL.md` * `skills/jira-create-release-note/SKILL.md` * `skills/jira-create-story/SKILL.md` * `skills/jira-create-task/SKILL.md` * `skills/jira-extract-prs/SKILL.md` * `skills/jira-gcp-hcp/CLAUDE.md` * `skills/jira-gcp-hcp/SKILL.md` * `skills/jira-hypershift/SKILL.md` * `skills/jira-jira-doc-generator/SKILL.md` * `skills/jira-jira-issues-by-component/README.md` * `skills/jira-jira-issues-by-component/SKILL.md` * `skills/jira-jira-issues-by-component/jira_curl.sh` * `skills/jira-jira-validate-blockers/SKILL.md` * `skills/jira-ocpbugs/SKILL.md` * `skills/jira-status-analysis/SKILL.md` * `skills/jira-status-analysis/activity-analysis.md` * `skills/jira-status-analysis/data-collection.md` * `skills/jira-status-analysis/external-links.md` * `skills/jira-status-analysis/formatting.md` * `skills/jira-status-analysis/scripts/gather_status_data.py` * `skills/lvms-lvms-analyzer/SKILL.md` * `skills/lvms-lvms-analyzer/scripts/analyze_lvms.py` * `skills/must-gather-must-gather-analyzer/SKILL.md` * `skills/must-gather-must-gather-analyzer/scripts/analyze_clusteroperators.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_clusterversion.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_etcd.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_events.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_network.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_nodes.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_ovn_dbs.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_pods.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_prometheus.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_pvs.py` * `skills/must-gather-must-gather-analyzer/scripts/analyze_windows_logs.py` * `skills/node-tuning-scripts/SKILL.md` * `skills/node-tuning-scripts/analyze_node_tuning.py` * `skills/node-tuning-scripts/generate_tuned_profile.py` * `skills/olm-team-k8s-ocp-olm-expert/README.md` * `skills/olm-team-k8s-ocp-olm-expert/SKILL.md` * `skills/olm-team-k8s-ocp-olm-expert/config-example.json` * `skills/olm-team-k8s-ocp-olm-expert/config-template.json` * `skills/openshift-generating-ovn-topology/README.md` * `skills/openshift-generating-ovn-topology/SKILL.md` * `skills/openshift-generating-ovn-topology/scripts/analyze_placement.py` * `skills/openshift-generating-ovn-topology/scripts/check_permissions.py` * `skills/openshift-generating-ovn-topology/scripts/collect_ovn_data.py` * `skills/openshift-generating-ovn-topology/scripts/detect-cluster.sh` * `skills/openshift-generating-ovn-topology/scripts/ovn_utils.py` * `skills/openshift-openshift-node-kernel/SKILL.md` * `skills/openshift-openshift-node-kernel/kernel-helper.sh` * `skills/openshift-openshift-node-kernel/node-kernel-conntrack.sh` * `skills/openshift-openshift-node-kernel/node-kernel-ip.sh` * `skills/openshift-openshift-node-kernel/node-kernel-iptables.sh` * `skills/openshift-openshift-node-kernel/node-kernel-nft.sh` * `skills/ote-migration-ote-migration-workflow/SKILL.md` * `skills/sosreport-logs-analysis/SKILL.md` * `skills/sosreport-network-analysis/SKILL.md` * `skills/sosreport-ovs-db-analysis/SKILL.md` * `skills/sosreport-ovs-db-analysis/scripts/analyze_ovs_db.py` * `skills/sosreport-resource-analysis/SKILL.md` * `skills/sosreport-system-config-analysis/SKILL.md` * `skills/teams-analyze-regressions/README.md` * `skills/teams-analyze-regressions/SKILL.md` * `skills/teams-analyze-regressions/generate_html_report.py` * `skills/teams-analyze-regressions/report_template.html` * `skills/teams-coderabbit-adoption/SKILL.md` * `skills/teams-coderabbit-adoption/coderabbit_adoption.py` * `skills/teams-coderabbit-inheritance-scanner-check/SKILL.md` * `skills/teams-coderabbit-inheritance-scanner-existing-pr/SKILL.md` * `skills/teams-coderabbit-inheritance-scanner-open-pr/SKILL.md` * `skills/teams-coderabbit-inheritance-scanner-search/SKILL.md` * `skills/teams-get-release-dates/README.md` * `skills/teams-get-release-dates/SKILL.md` * `skills/teams-get-release-dates/get_release_dates.py` * `skills/teams-list-components/SKILL.md` * `skills/teams-list-components/list_components.py` * `skills/teams-list-jiras/SKILL.md` * `skills/teams-list-jiras/list_jiras.py` * `skills/teams-list-regressions/README.md` * `skills/teams-list-regressions/SKILL.md` * `skills/teams-list-regressions/list_regressions.py` * `skills/teams-list-teams/SKILL.md` * `skills/teams-list-teams/list_teams.py` * `skills/teams-summarize-jiras/SKILL.md` </details> <details> <summary>✅ Files skipped from review due to trivial changes (1)</summary> * commands/container-image/compare.toml </details> <details> <summary>🚧 Files skipped from review as they are similar to previous changes (27)</summary> * commands/git/summary.toml * commands/hello-world/echo.toml * commands/container-image/inspect.toml * commands/container-image/tags.toml * commands/ci/query-job-status.toml * commands/code-review/pre-commit-review.toml * commands/ci/revert-pr.toml * commands/ci/ask-sippy.toml * commands/ci/analyze-prow-job-resource.toml * commands/ci/extract-prow-job-must-gather.toml * commands/jira/status-rollup.toml * commands/jira/generate-feature-doc.toml * commands/ci/query-test-result.toml * commands/agendas/outcome-refinement.toml * commands/git/cherry-pick-by-patch.toml * commands/git/commit-suggest.toml * commands/git/redescribe.toml * commands/hcp/cluster-health-check.toml * commands/gwapi/delete.toml * commands/jira/validate-blockers.toml * commands/ci/fetch-payloads.toml * commands/node-tuning/generate-tuned-profile.toml * GEMINI.md * commands/golang/lint-fix.toml * commands/git/fix-cherrypick-robot-pr.toml * commands/ci/list-unstable-tests.toml * commands/git/debt-scan.toml </details> </details> <!-- This is an auto-generated comment by CodeRabbit for review status -->
Contributor
|
PR is too big to review. better to split into many small ones one by one. |
Contributor
Author
|
Sorry, I think most of changes are added by the generated files. So not sure it can be seperated in another PR? |
Member
|
I would split out the code that does the conversion and the converison itself. CodeRabbit also seems intent on reviewing all the generated files, too. We should exclude anything generated: https://docs.coderabbit.ai/reference/configuration |
3 tasks
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 10, 2026 03:06
67c2b11 to
16e1603
Compare
Contributor
Author
|
Thanks @stbenjam - I split them to two seperated commits and add exclude the generated files from review in |
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
Makefile (1)
37-43: Reuseconvert-to-geminifromupdate.
updatenow duplicates the same conversion recipe asconvert-to-gemini, so future changes to the invocation can drift between the two targets.♻️ Proposed refactor
`@echo` "Building website data..." `@python3` scripts/build-website.py - `@echo` "Converting plugins to Gemini extensions..." - `@python3` scripts/convert_to_gemini.py + @$(MAKE) convert-to-gemini🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@Makefile` around lines 37 - 43, Duplicate conversion recipe exists in the Makefile: remove the duplicated commands under the update target and make update invoke the canonical convert-to-gemini target instead; specifically, replace the duplicated echo/python lines in the update target with a call to the convert-to-gemini target using $(MAKE) or make so update depends on and reuses convert-to-gemini (reference targets: convert-to-gemini and update).
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@Makefile`:
- Around line 45-47: The Makefile target verify-gemini-sync has a stale help
description referencing gemini-extensions/ but actually runs "python3
scripts/convert_to_gemini.py --check" to verify root-level generated Gemini
artifacts; update the target's comment/description line to accurately describe
that it verifies root-level generated Gemini artifacts (or "generated Gemini
files in the repository root") so contributors are pointed to the correct
location when the check fails—modify the comment next to the verify-gemini-sync
target to mention the root-level generated Gemini artifacts and/or the
scripts/convert_to_gemini.py --check check.
In `@README.md`:
- Around line 71-74: Replace the hardcoded remote name in the README example
command '/jira:solve OCPBUGS-12345 origin' with a placeholder (e.g., '<remote>')
and update the surrounding text to instruct readers to discover their remote
first (suggest using 'git remote -v' or 'git branch -vv'); modify the example
command and adjacent explanation so the README no longer assumes a remote named
"origin" and instead guides users to determine the correct remote for their fork
or repo.
---
Nitpick comments:
In `@Makefile`:
- Around line 37-43: Duplicate conversion recipe exists in the Makefile: remove
the duplicated commands under the update target and make update invoke the
canonical convert-to-gemini target instead; specifically, replace the duplicated
echo/python lines in the update target with a call to the convert-to-gemini
target using $(MAKE) or make so update depends on and reuses convert-to-gemini
(reference targets: convert-to-gemini and update).
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: ed93f5c9-c111-4b98-95ee-7bf011fa484e
⛔ Files ignored due to path filters (295)
GEMINI.mdis excluded by!GEMINI.mdcommands/agendas/outcome-refinement.tomlis excluded by!commands/**commands/bigquery/analyze-usage.tomlis excluded by!commands/**commands/ci/add-debug-wait.tomlis excluded by!commands/**commands/ci/analyze-payload.tomlis excluded by!commands/**commands/ci/analyze-pr-reverts.tomlis excluded by!commands/**commands/ci/analyze-prow-job-install-failure.tomlis excluded by!commands/**commands/ci/analyze-prow-job-resource.tomlis excluded by!commands/**commands/ci/analyze-prow-job-test-failure.tomlis excluded by!commands/**commands/ci/analyze-regression.tomlis excluded by!commands/**commands/ci/ask-sippy.tomlis excluded by!commands/**commands/ci/check-if-jira-regression-is-ongoing.tomlis excluded by!commands/**commands/ci/extract-prow-job-must-gather.tomlis excluded by!commands/**commands/ci/fetch-payloads.tomlis excluded by!commands/**commands/ci/fetch-test-report.tomlis excluded by!commands/**commands/ci/list-step.tomlis excluded by!commands/**commands/ci/list-unstable-tests.tomlis excluded by!commands/**commands/ci/payload-agent.tomlis excluded by!commands/**commands/ci/query-job-status.tomlis excluded by!commands/**commands/ci/query-test-result.tomlis excluded by!commands/**commands/ci/revert-pr.tomlis excluded by!commands/**commands/ci/trigger-periodic.tomlis excluded by!commands/**commands/ci/trigger-postsubmit.tomlis excluded by!commands/**commands/ci/trigger-presubmit.tomlis excluded by!commands/**commands/code-review/pr.tomlis excluded by!commands/**commands/code-review/pre-commit-review.tomlis excluded by!commands/**commands/compliance/analyze-cve.tomlis excluded by!commands/**commands/container-image/compare.tomlis excluded by!commands/**commands/container-image/inspect.tomlis excluded by!commands/**commands/container-image/tags.tomlis excluded by!commands/**commands/doc/note.tomlis excluded by!commands/**commands/etcd/analyze-performance.tomlis excluded by!commands/**commands/etcd/health-check.tomlis excluded by!commands/**commands/git/backport.tomlis excluded by!commands/**commands/git/bisect.tomlis excluded by!commands/**commands/git/branch-cleanup.tomlis excluded by!commands/**commands/git/cherry-pick-by-patch.tomlis excluded by!commands/**commands/git/commit-suggest.tomlis excluded by!commands/**commands/git/debt-scan.tomlis excluded by!commands/**commands/git/fix-cherrypick-robot-pr.tomlis excluded by!commands/**commands/git/redescribe.tomlis excluded by!commands/**commands/git/suggest-reviewers.tomlis excluded by!commands/**commands/git/summary.tomlis excluded by!commands/**commands/golang/lint-fix.tomlis excluded by!commands/**commands/gwapi/check.tomlis excluded by!commands/**commands/gwapi/delete.tomlis excluded by!commands/**commands/gwapi/install.tomlis excluded by!commands/**commands/hcp/cluster-health-check.tomlis excluded by!commands/**commands/hcp/generate.tomlis excluded by!commands/**commands/hello-world/echo.tomlis excluded by!commands/**commands/jira/backlog.tomlis excluded by!commands/**commands/jira/categorize-activity-type.tomlis excluded by!commands/**commands/jira/clone-from-github.tomlis excluded by!commands/**commands/jira/create-release-note.tomlis excluded by!commands/**commands/jira/create.tomlis excluded by!commands/**commands/jira/generate-feature-doc.tomlis excluded by!commands/**commands/jira/generate-test-plan.tomlis excluded by!commands/**commands/jira/grooming.tomlis excluded by!commands/**commands/jira/issues-by-component.tomlis excluded by!commands/**commands/jira/reconcile-github.tomlis excluded by!commands/**commands/jira/setup-gh2jira.tomlis excluded by!commands/**commands/jira/solve.tomlis excluded by!commands/**commands/jira/status-rollup.tomlis excluded by!commands/**commands/jira/update-weekly-status.tomlis excluded by!commands/**commands/jira/validate-blockers.tomlis excluded by!commands/**commands/lvms/analyze.tomlis excluded by!commands/**commands/must-gather/analyze.tomlis excluded by!commands/**commands/must-gather/ovn-dbs.tomlis excluded by!commands/**commands/must-gather/windows.tomlis excluded by!commands/**commands/node-tuning/analyze-node-tuning.tomlis excluded by!commands/**commands/node-tuning/generate-tuned-profile.tomlis excluded by!commands/**commands/node/cluster-node-health-check.tomlis excluded by!commands/**commands/olm-team/configure-agent.tomlis excluded by!commands/**commands/olm-team/dev-setup.tomlis excluded by!commands/**commands/olm-team/ep-watch.tomlis excluded by!commands/**commands/olm/approve.tomlis excluded by!commands/**commands/olm/catalog.tomlis excluded by!commands/**commands/olm/debug.tomlis excluded by!commands/**commands/olm/diagnose.tomlis excluded by!commands/**commands/olm/install.tomlis excluded by!commands/**commands/olm/list.tomlis excluded by!commands/**commands/olm/opm.tomlis excluded by!commands/**commands/olm/search.tomlis excluded by!commands/**commands/olm/status.tomlis excluded by!commands/**commands/olm/uninstall.tomlis excluded by!commands/**commands/olm/upgrade.tomlis excluded by!commands/**commands/openshift/add-enhancement.tomlis excluded by!commands/**commands/openshift/bootstrap-om.tomlis excluded by!commands/**commands/openshift/bump-deps.tomlis excluded by!commands/**commands/openshift/cluster-health-check.tomlis excluded by!commands/**commands/openshift/crd-review.tomlis excluded by!commands/**commands/openshift/create-cluster.tomlis excluded by!commands/**commands/openshift/destroy-cluster.tomlis excluded by!commands/**commands/openshift/expand-test-case.tomlis excluded by!commands/**commands/openshift/ironic-status.tomlis excluded by!commands/**commands/openshift/new-e2e-test.tomlis excluded by!commands/**commands/openshift/node-kernel-conntrack.tomlis excluded by!commands/**commands/openshift/node-kernel-ip.tomlis excluded by!commands/**commands/openshift/node-kernel-iptables.tomlis excluded by!commands/**commands/openshift/node-kernel-nft.tomlis excluded by!commands/**commands/openshift/rebase.tomlis excluded by!commands/**commands/openshift/review-test-cases.tomlis excluded by!commands/**commands/openshift/visualize-ovn-topology.tomlis excluded by!commands/**commands/origin/two-node-origin-pr-helper.tomlis excluded by!commands/**commands/ote-migration/migrate.tomlis excluded by!commands/**commands/session/save-session.tomlis excluded by!commands/**commands/sosreport/analyze.tomlis excluded by!commands/**commands/sosreport/ovs-db.tomlis excluded by!commands/**commands/teams/coderabbit-adoption-report.tomlis excluded by!commands/**commands/teams/coderabbit-inheritance-scanner.tomlis excluded by!commands/**commands/teams/health-check-jiras.tomlis excluded by!commands/**commands/teams/health-check-regressions.tomlis excluded by!commands/**commands/teams/health-check.tomlis excluded by!commands/**commands/teams/list-components.tomlis excluded by!commands/**commands/teams/list-jiras.tomlis excluded by!commands/**commands/teams/list-regressions.tomlis excluded by!commands/**commands/teams/list-teams.tomlis excluded by!commands/**commands/test-coverage/analyze.tomlis excluded by!commands/**commands/test-coverage/gaps.tomlis excluded by!commands/**commands/testing/mutation-test.tomlis excluded by!commands/**commands/utils/address-reviews.tomlis excluded by!commands/**commands/utils/auto-approve-konflux-prs.tomlis excluded by!commands/**commands/utils/generate-test-plan.tomlis excluded by!commands/**commands/utils/gh-attention.tomlis excluded by!commands/**commands/utils/placeholder.tomlis excluded by!commands/**commands/utils/process-renovate-pr.tomlis excluded by!commands/**commands/utils/review-ai-helpers-overlap.tomlis excluded by!commands/**commands/utils/review-security.tomlis excluded by!commands/**commands/workspaces/create.tomlis excluded by!commands/**commands/workspaces/delete.tomlis excluded by!commands/**commands/yaml/docs.tomlis excluded by!commands/**gemini-extension.jsonis excluded by!gemini-extension.jsonskills/bigquery-analyze-usage/SKILL.mdis excluded by!skills/**skills/ci-analyze-payload/SKILL.mdis excluded by!skills/**skills/ci-bisect-payload-suspects/SKILL.mdis excluded by!skills/**skills/ci-fetch-jira-issue/README.mdis excluded by!skills/**skills/ci-fetch-jira-issue/SKILL.mdis excluded by!skills/**skills/ci-fetch-jira-issue/fetch_jira_issue.pyis excluded by!skills/**skills/ci-fetch-new-prs-in-payload/SKILL.mdis excluded by!skills/**skills/ci-fetch-new-prs-in-payload/fetch_new_prs_in_payload.pyis excluded by!skills/**skills/ci-fetch-payloads/SKILL.mdis excluded by!skills/**skills/ci-fetch-payloads/fetch_payloads.pyis excluded by!skills/**skills/ci-fetch-prowjob-json/SKILL.mdis excluded by!skills/**skills/ci-fetch-regression-details/README.mdis excluded by!skills/**skills/ci-fetch-regression-details/SKILL.mdis excluded by!skills/**skills/ci-fetch-regression-details/fetch_regression_details.pyis excluded by!skills/**skills/ci-fetch-related-triages/SKILL.mdis excluded by!skills/**skills/ci-fetch-related-triages/fetch_related_triages.pyis excluded by!skills/**skills/ci-fetch-releases/SKILL.mdis excluded by!skills/**skills/ci-fetch-releases/fetch_releases.pyis excluded by!skills/**skills/ci-fetch-test-report/SKILL.mdis excluded by!skills/**skills/ci-fetch-test-report/fetch_test_report.pyis excluded by!skills/**skills/ci-fetch-test-runs/README.mdis excluded by!skills/**skills/ci-fetch-test-runs/SKILL.mdis excluded by!skills/**skills/ci-fetch-test-runs/fetch_test_runs.pyis excluded by!skills/**skills/ci-oc-auth/README.mdis excluded by!skills/**skills/ci-oc-auth/SKILL.mdis excluded by!skills/**skills/ci-oc-auth/curl_with_token.shis excluded by!skills/**skills/ci-payload-agent/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-install-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-metal-install-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/CHANGELOG.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/README.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/SCRIPTS.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/create_context_html_files.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/create_inline_html_files.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/generate_html_report.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/generate_report.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_all_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_audit_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_pod_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_url.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/prow_job_resource_grep.shis excluded by!skills/**skills/ci-prow-job-analyze-resource/report_template.htmlis excluded by!skills/**skills/ci-prow-job-analyze-test-failure/README.mdis excluded by!skills/**skills/ci-prow-job-analyze-test-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-artifact-search/SKILL.mdis excluded by!skills/**skills/ci-prow-job-artifact-search/prow_job_artifact_search.pyis excluded by!skills/**skills/ci-prow-job-extract-must-gather/CHANGELOG.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/README.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/SKILL.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/extract_archives.pyis excluded by!skills/**skills/ci-prow-job-extract-must-gather/generate_html_report.pyis excluded by!skills/**skills/ci-revert-pr/SKILL.mdis excluded by!skills/**skills/ci-set-release-blocker/SKILL.mdis excluded by!skills/**skills/ci-set-release-blocker/set_release_blocker.pyis excluded by!skills/**skills/ci-stage-payload-reverts/SKILL.mdis excluded by!skills/**skills/ci-triage-regression/README.mdis excluded by!skills/**skills/ci-triage-regression/SKILL.mdis excluded by!skills/**skills/ci-triage-regression/triage_regression.pyis excluded by!skills/**skills/ci-trigger-payload-job/SKILL.mdis excluded by!skills/**skills/code-review-lang-go/SKILL.mdis excluded by!skills/**skills/code-review-profile-hypershift/SKILL.mdis excluded by!skills/**skills/compliance-call-graph-analysis/SKILL.mdis excluded by!skills/**skills/compliance-codebase-impact-analysis/SKILL.mdis excluded by!skills/**skills/compliance-cve-intelligence-gathering/SKILL.mdis excluded by!skills/**skills/compliance-remediation-planning/SKILL.mdis excluded by!skills/**skills/git-suggest-reviewers/SKILL.mdis excluded by!skills/**skills/git-suggest-reviewers/analyze_blame.pyis excluded by!skills/**skills/golang-lint/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-agent/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-aws/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-azure/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-kubevirt/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-openstack/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-powervs/SKILL.mdis excluded by!skills/**skills/jira-categorize-activity-type/SKILL.mdis excluded by!skills/**skills/jira-cntrlplane/SKILL.mdis excluded by!skills/**skills/jira-create-bug/SKILL.mdis excluded by!skills/**skills/jira-create-epic/SKILL.mdis excluded by!skills/**skills/jira-create-feature-request/SKILL.mdis excluded by!skills/**skills/jira-create-feature/SKILL.mdis excluded by!skills/**skills/jira-create-release-note/SKILL.mdis excluded by!skills/**skills/jira-create-story/SKILL.mdis excluded by!skills/**skills/jira-create-task/SKILL.mdis excluded by!skills/**skills/jira-extract-prs/SKILL.mdis excluded by!skills/**skills/jira-gcp-hcp/CLAUDE.mdis excluded by!skills/**skills/jira-gcp-hcp/SKILL.mdis excluded by!skills/**skills/jira-hypershift/SKILL.mdis excluded by!skills/**skills/jira-jira-doc-generator/SKILL.mdis excluded by!skills/**skills/jira-jira-issues-by-component/README.mdis excluded by!skills/**skills/jira-jira-issues-by-component/SKILL.mdis excluded by!skills/**skills/jira-jira-issues-by-component/jira_curl.shis excluded by!skills/**skills/jira-jira-validate-blockers/SKILL.mdis excluded by!skills/**skills/jira-ocpbugs/SKILL.mdis excluded by!skills/**skills/jira-status-analysis/SKILL.mdis excluded by!skills/**skills/jira-status-analysis/activity-analysis.mdis excluded by!skills/**skills/jira-status-analysis/data-collection.mdis excluded by!skills/**skills/jira-status-analysis/external-links.mdis excluded by!skills/**skills/jira-status-analysis/formatting.mdis excluded by!skills/**skills/jira-status-analysis/scripts/gather_status_data.pyis excluded by!skills/**skills/lvms-lvms-analyzer/SKILL.mdis excluded by!skills/**skills/lvms-lvms-analyzer/scripts/analyze_lvms.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/SKILL.mdis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_clusteroperators.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_clusterversion.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_etcd.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_events.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_network.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_nodes.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_ovn_dbs.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_pods.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_prometheus.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_pvs.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_windows_logs.pyis excluded by!skills/**skills/node-tuning-scripts/SKILL.mdis excluded by!skills/**skills/node-tuning-scripts/analyze_node_tuning.pyis excluded by!skills/**skills/node-tuning-scripts/generate_tuned_profile.pyis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/README.mdis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/SKILL.mdis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/config-example.jsonis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/config-template.jsonis excluded by!skills/**skills/openshift-generating-ovn-topology/README.mdis excluded by!skills/**skills/openshift-generating-ovn-topology/SKILL.mdis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/analyze_placement.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/check_permissions.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/collect_ovn_data.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/detect-cluster.shis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/ovn_utils.pyis excluded by!skills/**skills/openshift-openshift-node-kernel/SKILL.mdis excluded by!skills/**skills/openshift-openshift-node-kernel/kernel-helper.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-conntrack.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-ip.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-iptables.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-nft.shis excluded by!skills/**skills/ote-migration-ote-migration-workflow/SKILL.mdis excluded by!skills/**skills/sosreport-logs-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-network-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-ovs-db-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-ovs-db-analysis/scripts/analyze_ovs_db.pyis excluded by!skills/**skills/sosreport-resource-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-system-config-analysis/SKILL.mdis excluded by!skills/**skills/teams-analyze-regressions/README.mdis excluded by!skills/**skills/teams-analyze-regressions/SKILL.mdis excluded by!skills/**skills/teams-analyze-regressions/generate_html_report.pyis excluded by!skills/**skills/teams-analyze-regressions/report_template.htmlis excluded by!skills/**skills/teams-coderabbit-adoption/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-adoption/coderabbit_adoption.pyis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-check/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-existing-pr/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-open-pr/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-search/SKILL.mdis excluded by!skills/**skills/teams-get-release-dates/README.mdis excluded by!skills/**skills/teams-get-release-dates/SKILL.mdis excluded by!skills/**skills/teams-get-release-dates/get_release_dates.pyis excluded by!skills/**skills/teams-list-components/SKILL.mdis excluded by!skills/**skills/teams-list-components/list_components.pyis excluded by!skills/**skills/teams-list-jiras/SKILL.mdis excluded by!skills/**skills/teams-list-jiras/list_jiras.pyis excluded by!skills/**skills/teams-list-regressions/README.mdis excluded by!skills/**skills/teams-list-regressions/SKILL.mdis excluded by!skills/**skills/teams-list-regressions/list_regressions.pyis excluded by!skills/**skills/teams-list-teams/SKILL.mdis excluded by!skills/**skills/teams-list-teams/list_teams.pyis excluded by!skills/**
📒 Files selected for processing (5)
.coderabbit.yaml.github/workflows/verify-gemini-sync.ymlMakefileREADME.mdscripts/convert_to_gemini.py
🚧 Files skipped from review as they are similar to previous changes (1)
- .github/workflows/verify-gemini-sync.yml
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 10, 2026 03:21
16e1603 to
d1cf0c3
Compare
Contributor
There was a problem hiding this comment.
♻️ Duplicate comments (1)
README.md (1)
71-74:⚠️ Potential issue | 🟠 MajorReplace the hardcoded remote and label the fenced block.
This example still assumes
origin, which is exactly the case the repo’s git-remote safety rules try to avoid, and Line 72 is missing a language identifier. Use a placeholder remote, tell readers to discover it first, and mark the block asbash.🛠️ Proposed change
-**Use the commands:** -``` -/jira:solve OCPBUGS-12345 origin -``` +**Use the commands:** +```bash +/jira:solve OCPBUGS-12345 <remote> +``` + +Discover the remote name first with `git remote -v` or `git branch -vv`.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@README.md` around lines 71 - 74, Replace the hardcoded remote in the example command "/jira:solve OCPBUGS-12345 origin" with a placeholder (e.g. "<remote>"), add the fenced block language identifier "bash" to the triple-backtick fence, and include a short note instructing readers to discover their remote name using commands like "git remote -v" or "git branch -vv" so they know what to substitute for <remote>; update the README.md example accordingly.
🧹 Nitpick comments (1)
.github/workflows/verify-gemini-sync.yml (1)
26-27: Prefer the Make target as the CI entrypoint.This duplicates the verification command instead of reusing
make verify-gemini-sync, so CI can drift from the documented local path over time.♻️ Proposed change
- - name: Verify Gemini extension is in sync - run: python3 scripts/convert_to_gemini.py --check + - name: Verify Gemini extension is in sync + run: make verify-gemini-sync🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.github/workflows/verify-gemini-sync.yml around lines 26 - 27, The CI step currently runs the verification script directly; replace that run command to invoke the Make target instead so the pipeline uses the canonical entrypoint (use the make target verify-gemini-sync rather than calling python3 scripts/convert_to_gemini.py --check). Update the job step that currently has the run command to call make verify-gemini-sync (ensuring the runner has required tools for the Make target), so CI and local verification remain in sync.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@README.md`:
- Around line 71-74: Replace the hardcoded remote in the example command
"/jira:solve OCPBUGS-12345 origin" with a placeholder (e.g. "<remote>"), add the
fenced block language identifier "bash" to the triple-backtick fence, and
include a short note instructing readers to discover their remote name using
commands like "git remote -v" or "git branch -vv" so they know what to
substitute for <remote>; update the README.md example accordingly.
---
Nitpick comments:
In @.github/workflows/verify-gemini-sync.yml:
- Around line 26-27: The CI step currently runs the verification script
directly; replace that run command to invoke the Make target instead so the
pipeline uses the canonical entrypoint (use the make target verify-gemini-sync
rather than calling python3 scripts/convert_to_gemini.py --check). Update the
job step that currently has the run command to call make verify-gemini-sync
(ensuring the runner has required tools for the Make target), so CI and local
verification remain in sync.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: a6037064-bee2-4248-a44d-5671d3f9babd
⛔ Files ignored due to path filters (295)
GEMINI.mdis excluded by!GEMINI.mdcommands/agendas/outcome-refinement.tomlis excluded by!commands/**commands/bigquery/analyze-usage.tomlis excluded by!commands/**commands/ci/add-debug-wait.tomlis excluded by!commands/**commands/ci/analyze-payload.tomlis excluded by!commands/**commands/ci/analyze-pr-reverts.tomlis excluded by!commands/**commands/ci/analyze-prow-job-install-failure.tomlis excluded by!commands/**commands/ci/analyze-prow-job-resource.tomlis excluded by!commands/**commands/ci/analyze-prow-job-test-failure.tomlis excluded by!commands/**commands/ci/analyze-regression.tomlis excluded by!commands/**commands/ci/ask-sippy.tomlis excluded by!commands/**commands/ci/check-if-jira-regression-is-ongoing.tomlis excluded by!commands/**commands/ci/extract-prow-job-must-gather.tomlis excluded by!commands/**commands/ci/fetch-payloads.tomlis excluded by!commands/**commands/ci/fetch-test-report.tomlis excluded by!commands/**commands/ci/list-step.tomlis excluded by!commands/**commands/ci/list-unstable-tests.tomlis excluded by!commands/**commands/ci/payload-agent.tomlis excluded by!commands/**commands/ci/query-job-status.tomlis excluded by!commands/**commands/ci/query-test-result.tomlis excluded by!commands/**commands/ci/revert-pr.tomlis excluded by!commands/**commands/ci/trigger-periodic.tomlis excluded by!commands/**commands/ci/trigger-postsubmit.tomlis excluded by!commands/**commands/ci/trigger-presubmit.tomlis excluded by!commands/**commands/code-review/pr.tomlis excluded by!commands/**commands/code-review/pre-commit-review.tomlis excluded by!commands/**commands/compliance/analyze-cve.tomlis excluded by!commands/**commands/container-image/compare.tomlis excluded by!commands/**commands/container-image/inspect.tomlis excluded by!commands/**commands/container-image/tags.tomlis excluded by!commands/**commands/doc/note.tomlis excluded by!commands/**commands/etcd/analyze-performance.tomlis excluded by!commands/**commands/etcd/health-check.tomlis excluded by!commands/**commands/git/backport.tomlis excluded by!commands/**commands/git/bisect.tomlis excluded by!commands/**commands/git/branch-cleanup.tomlis excluded by!commands/**commands/git/cherry-pick-by-patch.tomlis excluded by!commands/**commands/git/commit-suggest.tomlis excluded by!commands/**commands/git/debt-scan.tomlis excluded by!commands/**commands/git/fix-cherrypick-robot-pr.tomlis excluded by!commands/**commands/git/redescribe.tomlis excluded by!commands/**commands/git/suggest-reviewers.tomlis excluded by!commands/**commands/git/summary.tomlis excluded by!commands/**commands/golang/lint-fix.tomlis excluded by!commands/**commands/gwapi/check.tomlis excluded by!commands/**commands/gwapi/delete.tomlis excluded by!commands/**commands/gwapi/install.tomlis excluded by!commands/**commands/hcp/cluster-health-check.tomlis excluded by!commands/**commands/hcp/generate.tomlis excluded by!commands/**commands/hello-world/echo.tomlis excluded by!commands/**commands/jira/backlog.tomlis excluded by!commands/**commands/jira/categorize-activity-type.tomlis excluded by!commands/**commands/jira/clone-from-github.tomlis excluded by!commands/**commands/jira/create-release-note.tomlis excluded by!commands/**commands/jira/create.tomlis excluded by!commands/**commands/jira/generate-feature-doc.tomlis excluded by!commands/**commands/jira/generate-test-plan.tomlis excluded by!commands/**commands/jira/grooming.tomlis excluded by!commands/**commands/jira/issues-by-component.tomlis excluded by!commands/**commands/jira/reconcile-github.tomlis excluded by!commands/**commands/jira/setup-gh2jira.tomlis excluded by!commands/**commands/jira/solve.tomlis excluded by!commands/**commands/jira/status-rollup.tomlis excluded by!commands/**commands/jira/update-weekly-status.tomlis excluded by!commands/**commands/jira/validate-blockers.tomlis excluded by!commands/**commands/lvms/analyze.tomlis excluded by!commands/**commands/must-gather/analyze.tomlis excluded by!commands/**commands/must-gather/ovn-dbs.tomlis excluded by!commands/**commands/must-gather/windows.tomlis excluded by!commands/**commands/node-tuning/analyze-node-tuning.tomlis excluded by!commands/**commands/node-tuning/generate-tuned-profile.tomlis excluded by!commands/**commands/node/cluster-node-health-check.tomlis excluded by!commands/**commands/olm-team/configure-agent.tomlis excluded by!commands/**commands/olm-team/dev-setup.tomlis excluded by!commands/**commands/olm-team/ep-watch.tomlis excluded by!commands/**commands/olm/approve.tomlis excluded by!commands/**commands/olm/catalog.tomlis excluded by!commands/**commands/olm/debug.tomlis excluded by!commands/**commands/olm/diagnose.tomlis excluded by!commands/**commands/olm/install.tomlis excluded by!commands/**commands/olm/list.tomlis excluded by!commands/**commands/olm/opm.tomlis excluded by!commands/**commands/olm/search.tomlis excluded by!commands/**commands/olm/status.tomlis excluded by!commands/**commands/olm/uninstall.tomlis excluded by!commands/**commands/olm/upgrade.tomlis excluded by!commands/**commands/openshift/add-enhancement.tomlis excluded by!commands/**commands/openshift/bootstrap-om.tomlis excluded by!commands/**commands/openshift/bump-deps.tomlis excluded by!commands/**commands/openshift/cluster-health-check.tomlis excluded by!commands/**commands/openshift/crd-review.tomlis excluded by!commands/**commands/openshift/create-cluster.tomlis excluded by!commands/**commands/openshift/destroy-cluster.tomlis excluded by!commands/**commands/openshift/expand-test-case.tomlis excluded by!commands/**commands/openshift/ironic-status.tomlis excluded by!commands/**commands/openshift/new-e2e-test.tomlis excluded by!commands/**commands/openshift/node-kernel-conntrack.tomlis excluded by!commands/**commands/openshift/node-kernel-ip.tomlis excluded by!commands/**commands/openshift/node-kernel-iptables.tomlis excluded by!commands/**commands/openshift/node-kernel-nft.tomlis excluded by!commands/**commands/openshift/rebase.tomlis excluded by!commands/**commands/openshift/review-test-cases.tomlis excluded by!commands/**commands/openshift/visualize-ovn-topology.tomlis excluded by!commands/**commands/origin/two-node-origin-pr-helper.tomlis excluded by!commands/**commands/ote-migration/migrate.tomlis excluded by!commands/**commands/session/save-session.tomlis excluded by!commands/**commands/sosreport/analyze.tomlis excluded by!commands/**commands/sosreport/ovs-db.tomlis excluded by!commands/**commands/teams/coderabbit-adoption-report.tomlis excluded by!commands/**commands/teams/coderabbit-inheritance-scanner.tomlis excluded by!commands/**commands/teams/health-check-jiras.tomlis excluded by!commands/**commands/teams/health-check-regressions.tomlis excluded by!commands/**commands/teams/health-check.tomlis excluded by!commands/**commands/teams/list-components.tomlis excluded by!commands/**commands/teams/list-jiras.tomlis excluded by!commands/**commands/teams/list-regressions.tomlis excluded by!commands/**commands/teams/list-teams.tomlis excluded by!commands/**commands/test-coverage/analyze.tomlis excluded by!commands/**commands/test-coverage/gaps.tomlis excluded by!commands/**commands/testing/mutation-test.tomlis excluded by!commands/**commands/utils/address-reviews.tomlis excluded by!commands/**commands/utils/auto-approve-konflux-prs.tomlis excluded by!commands/**commands/utils/generate-test-plan.tomlis excluded by!commands/**commands/utils/gh-attention.tomlis excluded by!commands/**commands/utils/placeholder.tomlis excluded by!commands/**commands/utils/process-renovate-pr.tomlis excluded by!commands/**commands/utils/review-ai-helpers-overlap.tomlis excluded by!commands/**commands/utils/review-security.tomlis excluded by!commands/**commands/workspaces/create.tomlis excluded by!commands/**commands/workspaces/delete.tomlis excluded by!commands/**commands/yaml/docs.tomlis excluded by!commands/**gemini-extension.jsonis excluded by!gemini-extension.jsonskills/bigquery-analyze-usage/SKILL.mdis excluded by!skills/**skills/ci-analyze-payload/SKILL.mdis excluded by!skills/**skills/ci-bisect-payload-suspects/SKILL.mdis excluded by!skills/**skills/ci-fetch-jira-issue/README.mdis excluded by!skills/**skills/ci-fetch-jira-issue/SKILL.mdis excluded by!skills/**skills/ci-fetch-jira-issue/fetch_jira_issue.pyis excluded by!skills/**skills/ci-fetch-new-prs-in-payload/SKILL.mdis excluded by!skills/**skills/ci-fetch-new-prs-in-payload/fetch_new_prs_in_payload.pyis excluded by!skills/**skills/ci-fetch-payloads/SKILL.mdis excluded by!skills/**skills/ci-fetch-payloads/fetch_payloads.pyis excluded by!skills/**skills/ci-fetch-prowjob-json/SKILL.mdis excluded by!skills/**skills/ci-fetch-regression-details/README.mdis excluded by!skills/**skills/ci-fetch-regression-details/SKILL.mdis excluded by!skills/**skills/ci-fetch-regression-details/fetch_regression_details.pyis excluded by!skills/**skills/ci-fetch-related-triages/SKILL.mdis excluded by!skills/**skills/ci-fetch-related-triages/fetch_related_triages.pyis excluded by!skills/**skills/ci-fetch-releases/SKILL.mdis excluded by!skills/**skills/ci-fetch-releases/fetch_releases.pyis excluded by!skills/**skills/ci-fetch-test-report/SKILL.mdis excluded by!skills/**skills/ci-fetch-test-report/fetch_test_report.pyis excluded by!skills/**skills/ci-fetch-test-runs/README.mdis excluded by!skills/**skills/ci-fetch-test-runs/SKILL.mdis excluded by!skills/**skills/ci-fetch-test-runs/fetch_test_runs.pyis excluded by!skills/**skills/ci-oc-auth/README.mdis excluded by!skills/**skills/ci-oc-auth/SKILL.mdis excluded by!skills/**skills/ci-oc-auth/curl_with_token.shis excluded by!skills/**skills/ci-payload-agent/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-install-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-metal-install-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/CHANGELOG.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/README.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/SCRIPTS.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/SKILL.mdis excluded by!skills/**skills/ci-prow-job-analyze-resource/create_context_html_files.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/create_inline_html_files.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/generate_html_report.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/generate_report.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_all_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_audit_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_pod_logs.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/parse_url.pyis excluded by!skills/**skills/ci-prow-job-analyze-resource/prow_job_resource_grep.shis excluded by!skills/**skills/ci-prow-job-analyze-resource/report_template.htmlis excluded by!skills/**skills/ci-prow-job-analyze-test-failure/README.mdis excluded by!skills/**skills/ci-prow-job-analyze-test-failure/SKILL.mdis excluded by!skills/**skills/ci-prow-job-artifact-search/SKILL.mdis excluded by!skills/**skills/ci-prow-job-artifact-search/prow_job_artifact_search.pyis excluded by!skills/**skills/ci-prow-job-extract-must-gather/CHANGELOG.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/README.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/SKILL.mdis excluded by!skills/**skills/ci-prow-job-extract-must-gather/extract_archives.pyis excluded by!skills/**skills/ci-prow-job-extract-must-gather/generate_html_report.pyis excluded by!skills/**skills/ci-revert-pr/SKILL.mdis excluded by!skills/**skills/ci-set-release-blocker/SKILL.mdis excluded by!skills/**skills/ci-set-release-blocker/set_release_blocker.pyis excluded by!skills/**skills/ci-stage-payload-reverts/SKILL.mdis excluded by!skills/**skills/ci-triage-regression/README.mdis excluded by!skills/**skills/ci-triage-regression/SKILL.mdis excluded by!skills/**skills/ci-triage-regression/triage_regression.pyis excluded by!skills/**skills/ci-trigger-payload-job/SKILL.mdis excluded by!skills/**skills/code-review-lang-go/SKILL.mdis excluded by!skills/**skills/code-review-profile-hypershift/SKILL.mdis excluded by!skills/**skills/compliance-call-graph-analysis/SKILL.mdis excluded by!skills/**skills/compliance-codebase-impact-analysis/SKILL.mdis excluded by!skills/**skills/compliance-cve-intelligence-gathering/SKILL.mdis excluded by!skills/**skills/compliance-remediation-planning/SKILL.mdis excluded by!skills/**skills/git-suggest-reviewers/SKILL.mdis excluded by!skills/**skills/git-suggest-reviewers/analyze_blame.pyis excluded by!skills/**skills/golang-lint/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-agent/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-aws/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-azure/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-kubevirt/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-openstack/SKILL.mdis excluded by!skills/**skills/hcp-hcp-create-powervs/SKILL.mdis excluded by!skills/**skills/jira-categorize-activity-type/SKILL.mdis excluded by!skills/**skills/jira-cntrlplane/SKILL.mdis excluded by!skills/**skills/jira-create-bug/SKILL.mdis excluded by!skills/**skills/jira-create-epic/SKILL.mdis excluded by!skills/**skills/jira-create-feature-request/SKILL.mdis excluded by!skills/**skills/jira-create-feature/SKILL.mdis excluded by!skills/**skills/jira-create-release-note/SKILL.mdis excluded by!skills/**skills/jira-create-story/SKILL.mdis excluded by!skills/**skills/jira-create-task/SKILL.mdis excluded by!skills/**skills/jira-extract-prs/SKILL.mdis excluded by!skills/**skills/jira-gcp-hcp/CLAUDE.mdis excluded by!skills/**skills/jira-gcp-hcp/SKILL.mdis excluded by!skills/**skills/jira-hypershift/SKILL.mdis excluded by!skills/**skills/jira-jira-doc-generator/SKILL.mdis excluded by!skills/**skills/jira-jira-issues-by-component/README.mdis excluded by!skills/**skills/jira-jira-issues-by-component/SKILL.mdis excluded by!skills/**skills/jira-jira-issues-by-component/jira_curl.shis excluded by!skills/**skills/jira-jira-validate-blockers/SKILL.mdis excluded by!skills/**skills/jira-ocpbugs/SKILL.mdis excluded by!skills/**skills/jira-status-analysis/SKILL.mdis excluded by!skills/**skills/jira-status-analysis/activity-analysis.mdis excluded by!skills/**skills/jira-status-analysis/data-collection.mdis excluded by!skills/**skills/jira-status-analysis/external-links.mdis excluded by!skills/**skills/jira-status-analysis/formatting.mdis excluded by!skills/**skills/jira-status-analysis/scripts/gather_status_data.pyis excluded by!skills/**skills/lvms-lvms-analyzer/SKILL.mdis excluded by!skills/**skills/lvms-lvms-analyzer/scripts/analyze_lvms.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/SKILL.mdis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_clusteroperators.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_clusterversion.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_etcd.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_events.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_network.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_nodes.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_ovn_dbs.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_pods.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_prometheus.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_pvs.pyis excluded by!skills/**skills/must-gather-must-gather-analyzer/scripts/analyze_windows_logs.pyis excluded by!skills/**skills/node-tuning-scripts/SKILL.mdis excluded by!skills/**skills/node-tuning-scripts/analyze_node_tuning.pyis excluded by!skills/**skills/node-tuning-scripts/generate_tuned_profile.pyis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/README.mdis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/SKILL.mdis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/config-example.jsonis excluded by!skills/**skills/olm-team-k8s-ocp-olm-expert/config-template.jsonis excluded by!skills/**skills/openshift-generating-ovn-topology/README.mdis excluded by!skills/**skills/openshift-generating-ovn-topology/SKILL.mdis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/analyze_placement.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/check_permissions.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/collect_ovn_data.pyis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/detect-cluster.shis excluded by!skills/**skills/openshift-generating-ovn-topology/scripts/ovn_utils.pyis excluded by!skills/**skills/openshift-openshift-node-kernel/SKILL.mdis excluded by!skills/**skills/openshift-openshift-node-kernel/kernel-helper.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-conntrack.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-ip.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-iptables.shis excluded by!skills/**skills/openshift-openshift-node-kernel/node-kernel-nft.shis excluded by!skills/**skills/ote-migration-ote-migration-workflow/SKILL.mdis excluded by!skills/**skills/sosreport-logs-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-network-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-ovs-db-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-ovs-db-analysis/scripts/analyze_ovs_db.pyis excluded by!skills/**skills/sosreport-resource-analysis/SKILL.mdis excluded by!skills/**skills/sosreport-system-config-analysis/SKILL.mdis excluded by!skills/**skills/teams-analyze-regressions/README.mdis excluded by!skills/**skills/teams-analyze-regressions/SKILL.mdis excluded by!skills/**skills/teams-analyze-regressions/generate_html_report.pyis excluded by!skills/**skills/teams-analyze-regressions/report_template.htmlis excluded by!skills/**skills/teams-coderabbit-adoption/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-adoption/coderabbit_adoption.pyis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-check/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-existing-pr/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-open-pr/SKILL.mdis excluded by!skills/**skills/teams-coderabbit-inheritance-scanner-search/SKILL.mdis excluded by!skills/**skills/teams-get-release-dates/README.mdis excluded by!skills/**skills/teams-get-release-dates/SKILL.mdis excluded by!skills/**skills/teams-get-release-dates/get_release_dates.pyis excluded by!skills/**skills/teams-list-components/SKILL.mdis excluded by!skills/**skills/teams-list-components/list_components.pyis excluded by!skills/**skills/teams-list-jiras/SKILL.mdis excluded by!skills/**skills/teams-list-jiras/list_jiras.pyis excluded by!skills/**skills/teams-list-regressions/README.mdis excluded by!skills/**skills/teams-list-regressions/SKILL.mdis excluded by!skills/**skills/teams-list-regressions/list_regressions.pyis excluded by!skills/**skills/teams-list-teams/SKILL.mdis excluded by!skills/**skills/teams-list-teams/list_teams.pyis excluded by!skills/**
📒 Files selected for processing (5)
.coderabbit.yaml.github/workflows/verify-gemini-sync.ymlMakefileREADME.mdscripts/convert_to_gemini.py
Contributor
|
My previous approach was to submit one or several small plugins as a PR, which was conducive to review and maintenance. See https://github.com/wangke19/gemini-ai-helpers/commits/main/ |
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 10, 2026 06:56
d1cf0c3 to
af9ef7e
Compare
Contributor
Author
|
Thanksk @wangke19 - now I removed all the generated gemini files and let's try to do the migration in more small steps. |
Contributor
There was a problem hiding this comment.
Actionable comments posted: 2
♻️ Duplicate comments (1)
README.md (1)
72-74:⚠️ Potential issue | 🟡 MinorAdd language specifier to fenced code block and avoid hardcoded remote name.
The code block at line 72 is missing a language specifier. Additionally, hardcoding
originhas been flagged in a previous review - this applies equally to the Gemini CLI example.📝 Proposed fix
**Use the commands:** -``` -/jira:solve OCPBUGS-12345 origin +```bash +/jira:solve OCPBUGS-12345 <remote>
+Discover the remote name with
git remote -v.</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@README.mdaround lines 72 - 74, Update the fenced code block containing the
example command "/jira:solve OCPBUGS-12345 origin" to include a bash language
specifier and remove the hardcoded remote name; replace the trailing "origin"
with a placeholder like "" and mirror this change in the Gemini CLI
example so both examples use the placeholder instead of a fixed remote; add a
short note suggesting users discover their remote name with "git remote -v".</details> </blockquote></details> </blockquote></details> <details> <summary>🧹 Nitpick comments (3)</summary><blockquote> <details> <summary>scripts/convert_to_gemini.py (3)</summary><blockquote> `614-615`: **Inconsistent temp directory cleanup handling.** Line 615 uses `shutil.rmtree(temp_dir)` without `ignore_errors`, while line 645 uses `ignore_errors=True`. If cleanup fails in `--check` mode, the script will crash after completing its work. Consider using `ignore_errors=True` consistently for both. <details> <summary>♻️ Proposed fix</summary> ```diff finally: - shutil.rmtree(temp_dir) + shutil.rmtree(temp_dir, ignore_errors=True)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@scripts/convert_to_gemini.py` around lines 614 - 615, The finally block currently calls shutil.rmtree(temp_dir) which can raise on failure and crash the script in --check mode; change that call inside the finally to shutil.rmtree(temp_dir, ignore_errors=True) to match the other cleanup at line 645 and ensure consistent, non-fatal temp directory cleanup for the function handling the conversion (referencing the temp_dir variable used in that try/finally).
100-105: Useraise ... from excfor proper exception chaining.Per static analysis (B904), re-raising with
from excpreserves the original traceback and distinguishes the new exception from errors in exception handling.♻️ Proposed fix
try: tomllib.loads(toml_content) except Exception as exc: - raise ValueError( + raise ValueError( f"Generated invalid TOML for {plugin_name}/{os.path.basename(md_path)}: {exc}" - ) + ) from exc🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@scripts/convert_to_gemini.py` around lines 100 - 105, In the except block that catches Exception as exc around tomllib.loads(toml_content), re-raise the ValueError using exception chaining so the original traceback is preserved: replace the current raise ValueError(...) with raise ValueError(f"Generated invalid TOML for {plugin_name}/{os.path.basename(md_path)}: {exc}") from exc (i.e., add "from exc" to the raise in the except handling of tomllib.loads).
636-636: Remove extraneousfprefix from string literal.The f-string has no placeholders.
♻️ Proposed fix
- print(f"Changed plugins:") + print("Changed plugins:")🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@scripts/convert_to_gemini.py` at line 636, The print statement uses an unnecessary f-string: replace the call print(f"Changed plugins:") with a normal string literal (print("Changed plugins:")) to remove the extraneous `f` prefix; locate the print(f"Changed plugins:") occurrence in convert_to_gemini.py and update it accordingly.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@scripts/convert_to_gemini.py`:
- Around line 333-334: The file comparison uses open(new_path).read() and
open(existing_path).read() which leaves file descriptors open; change these to
use context managers (with open(new_path, "r", encoding="utf-8") as f: data_new
= f.read()) and similarly for existing_path, then compare the strings (e.g.,
data_new != data_existing); apply the same context-manager fix for the other
occurrences referenced (the comparisons using open() at the other locations) to
ensure all file handles are properly closed and use explicit encodings.
- Line 96: The current line building the TOML prompt uses lines.append(f"prompt
= '''\n{body}'''") which will produce invalid TOML if body contains '''; update
the logic in scripts/convert_to_gemini.py to detect whether body contains triple
single quotes and, if so, either switch to a triple-double-quoted TOML multiline
string (e.g. use """...""" surrounding the body) or escape occurrences of '''
inside body (replace "'''" with "\\'\\'\\'") before appending; change the code
path that writes the prompt (the lines.append call referencing body) to choose
the safe quoting/escaping strategy based on the presence of '''.
---
Duplicate comments:
In `@README.md`:
- Around line 72-74: Update the fenced code block containing the example command
"/jira:solve OCPBUGS-12345 origin" to include a bash language specifier and
remove the hardcoded remote name; replace the trailing "origin" with a
placeholder like "<remote>" and mirror this change in the Gemini CLI example so
both examples use the placeholder instead of a fixed remote; add a short note
suggesting users discover their remote name with "git remote -v".
---
Nitpick comments:
In `@scripts/convert_to_gemini.py`:
- Around line 614-615: The finally block currently calls shutil.rmtree(temp_dir)
which can raise on failure and crash the script in --check mode; change that
call inside the finally to shutil.rmtree(temp_dir, ignore_errors=True) to match
the other cleanup at line 645 and ensure consistent, non-fatal temp directory
cleanup for the function handling the conversion (referencing the temp_dir
variable used in that try/finally).
- Around line 100-105: In the except block that catches Exception as exc around
tomllib.loads(toml_content), re-raise the ValueError using exception chaining so
the original traceback is preserved: replace the current raise ValueError(...)
with raise ValueError(f"Generated invalid TOML for
{plugin_name}/{os.path.basename(md_path)}: {exc}") from exc (i.e., add "from
exc" to the raise in the except handling of tomllib.loads).
- Line 636: The print statement uses an unnecessary f-string: replace the call
print(f"Changed plugins:") with a normal string literal (print("Changed
plugins:")) to remove the extraneous `f` prefix; locate the print(f"Changed
plugins:") occurrence in convert_to_gemini.py and update it accordingly.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 6e002d16-c1bd-4455-8214-b43dcc27d06a
📒 Files selected for processing (4)
.coderabbit.yamlMakefileREADME.mdscripts/convert_to_gemini.py
🚧 Files skipped from review as they are similar to previous changes (1)
- .coderabbit.yaml
| if description: | ||
| escaped_desc = description.replace("\\", "\\\\").replace('"', '\\"') | ||
| lines.append(f'description = "{escaped_desc}"') | ||
| lines.append(f"prompt = '''\n{body}'''") |
Contributor
There was a problem hiding this comment.
Triple single quotes in prompt body could break TOML parsing.
If a command's markdown body contains ''', the generated TOML will be invalid. Consider escaping or using a different quoting strategy.
🛡️ Proposed fix to handle edge case
- lines.append(f"prompt = '''\n{body}'''")
+ # Escape any ''' sequences in body to prevent TOML breakage
+ escaped_body = body.replace("'''", "'''\"'''\"'''")
+ lines.append(f"prompt = '''\n{escaped_body}'''")Alternatively, check if ''' exists and use double-quoted multiline string with proper escaping instead.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@scripts/convert_to_gemini.py` at line 96, The current line building the TOML
prompt uses lines.append(f"prompt = '''\n{body}'''") which will produce invalid
TOML if body contains '''; update the logic in scripts/convert_to_gemini.py to
detect whether body contains triple single quotes and, if so, either switch to a
triple-double-quoted TOML multiline string (e.g. use """...""" surrounding the
body) or escape occurrences of ''' inside body (replace "'''" with "\\'\\'\\'")
before appending; change the code path that writes the prompt (the lines.append
call referencing body) to choose the safe quoting/escaping strategy based on the
presence of '''.
Comment on lines
+333
to
+334
| if open(new_path).read() != open(existing_path).read(): | ||
| return True |
Contributor
There was a problem hiding this comment.
File handles not properly closed - potential resource leak.
Using open() without a context manager leaves file handles open until garbage collection. This pattern repeats at lines 390 and 404.
🛡️ Proposed fix
- if open(new_path).read() != open(existing_path).read():
+ with open(new_path) as f1, open(existing_path) as f2:
+ if f1.read() != f2.read():Apply similar fixes at lines 390 and 404.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| if open(new_path).read() != open(existing_path).read(): | |
| return True | |
| with open(new_path) as f1, open(existing_path) as f2: | |
| if f1.read() != f2.read(): | |
| return True |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@scripts/convert_to_gemini.py` around lines 333 - 334, The file comparison
uses open(new_path).read() and open(existing_path).read() which leaves file
descriptors open; change these to use context managers (with open(new_path, "r",
encoding="utf-8") as f: data_new = f.read()) and similarly for existing_path,
then compare the strings (e.g., data_new != data_existing); apply the same
context-manager fix for the other occurrences referenced (the comparisons using
open() at the other locations) to ensure all file handles are properly closed
and use explicit encodings.
Contributor
Author
Contributor
Code ReviewOverviewThis PR adds a Python script ( ✅ Strengths
🔴 Issues1. No Example of Generated Output in PRThere's no
2.
|
| Category | Assessment |
|---|---|
| Correctness | Mostly correct, but fragile YAML parsing could silently mangle frontmatter |
| Code quality | Good structure; significant duplication between convert_single_plugin and get_changed_plugins |
| Conventions | Follows existing script style; tomli dependency not declared |
| Test coverage | |
| Security | Low risk (read-only from plugins, write to repo root) |
The most important gap is #1: without committed generated output (or at least a sample), it's impossible to verify the extension actually loads in Gemini CLI. Running make convert-to-gemini and committing the result should be a prerequisite for merge.
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 11, 2026 04:23
af9ef7e to
a4a0aec
Compare
Contributor
Author
|
@wangke19 I address the major issues of
And I run the I think other ones could be improved later if needed. WDYT? |
Contributor
|
Here is a focused review of potential issues and architectural "hallucinations" to watch out for.
The Issue: Because the script iterates through the list and performs replacements on the same string, an earlier replacement could create a pattern that a later replacement accidentally triggers. Example: If you had a replacement ("Claude", "Gemini") and later ("Gemini-Extension", "New-Format"), a string that was originally "Claude-Extension" would be turned into "Gemini-Extension" and then transformed again. Fix: Use a single-pass regular expression with a mapping dictionary to ensure each segment of text is only replaced once.
The Risk: This will break if a value contains a colon (e.g., summary: "Task: Fix Bug") or if the frontmatter uses multi-line strings or comments. Claude Code Bias: Claude tends to write "good enough" parsers for standard cases but ignores the full YAML spec. If your plugins have complex metadata, this script will lose data.
Python The Fix: Use the same triple-single-quote ''' syntax for the description field as used for the prompt field.
Python The Risk: If the script is interrupted (or a disk error occurs) between the rmtree and the copytree, you lose your existing commands/skills entirely. The Fix: Use an "atomic" approach: copy to a .tmp folder next to the destination, then use os.replace() to swap them instantly.
The Observation: This means the global extension version could jump from 1.0.1 to 1.0.2 just by running the script on a single plugin, even if the other 10 plugins stayed the same. This is generally fine, but ensure the gemini-extension.json is intended to be a "Monorepo" version for all tools combined. |
Contributor
|
Review of convert-to-gemini Target
Suggestion: Add a check for requirements or a note in the help text if a virtual environment is expected.
Observation: If the Gemini extension is intended to be a first-class citizen in this repo, you should consider whether make update should automatically trigger make convert-to-gemini. Otherwise, a developer might update a Claude plugin, run make update, and forget to regenerate the Gemini version, leading to out-of-sync files in a PR.
Risk: If scripts/convert_to_gemini.py fails (e.g., due to a TOML validation error inside the script), make will technically report a failure, but since it's the last command, it's fine. However, it's good practice to ensure the script's exit codes are respected (which @ does, but double-check your script exits with sys.exit(1) on errors). Suggested Improvements B. Enhancing the Help Text |
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 11, 2026 16:27
a4a0aec to
20553dd
Compare
Contributor
Author
|
Thanks @wangke19 - now I remove all other changes and let's focus on the convert script now. All of the issues have been addressed. |
Contributor
|
The script looks good to me.
Step-by-step Step 1: install deps & convert hello-worldpip3 install pyyaml tomli Step 2: install the generated extension into Gemini CLIGemini CLI looks for extensions in ~/.gemini/extensions/mkdir -p ~/.gemini/extensions/ai-helpers Step 3: launch Gemini CLI and testgemini then type: /hello-world:echoWhat to verify
|
Member
|
/approve Feel free to lgtm when its ready. Looks reasonable to me. Does it make one extension with all plugins? I could see that causing issues with context but we can see how it goes |
openshift-ci
Bot
added
the
approved
Add convert_to_gemini.py that converts Claude Code plugins into a
Gemini CLI extension. Supports full conversion, single-plugin mode
(--plugin), and sync checking (--check).
Key features:
- Single-pass regex for Claude->Gemini text replacement
- YAML frontmatter parsed via yaml.safe_load
- Skills nested under skills/{plugin}/{skill}/ preserving paths
- ${CLAUDE_PLUGIN_ROOT} mapped to ${extensionPath}
- Atomic sync to avoid data loss on interruption
- Version auto-bumped only when content changes
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
zhfeng
force-pushed
the
feat/gemini-extension-conversion
branch
from
March 13, 2026 02:04
20553dd to
d927094
Compare
Contributor
Author
|
Hi all, I test with gemini locally
ln -s "$(pwd)" ~/.gemini/extensions/ai-helpers
python3 scripts/convert_to_gemini.py
I also some other commnds on git and ci. All of them looks good. So I'm happy to go for the next step and start to imigrate the plugins. |
Contributor
|
/lgtm |
Contributor
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: stbenjam, wangke19, zhfeng The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Contributor
Author
|
Hi @stbenjam, The first migration of |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add
scripts/convert_to_gemini.pyto convert Claude Code plugins into a Gemini CLI extension.yaml.safe_load(handles colons, multi-line values)'''syntax (handles newlines, special chars)skills/{plugin}/{skill}/to preserve path structure${CLAUDE_PLUGIN_ROOT}mapped to${extensionPath}with correct skill path rewritingadapt_text()applied to both command prompts and skill.mdfiles.tmp+ rename to avoid data loss on interruption--checkmode for CI gatingUsage
Text replacements
Claude Code pluginGemini CLI extensionClaude CodeGemini CLIClaudeGeminihttps://claude.com/claude-codehttps://github.com/google-gemini/gemini-cli~/.config/claude-code/~/.gemini/CLAUDE.mdGEMINI.md.claude/.gemini/claude-codegemini-cliclaudegemini${CLAUDE_PLUGIN_ROOT}/skills/X${extensionPath}/skills/{plugin}/X${CLAUDE_PLUGIN_ROOT}${extensionPath}Test plan
--plugin gitthen--plugin cibumps version from 1.0.0 to 1.0.1--plugin gittwice detects no changes, skips version bump--checkmode passes after full conversion${CLAUDE_PLUGIN_ROOT}/skills/suggest-reviewers/correctly maps to${extensionPath}/skills/git/suggest-reviewers/CLAUDE_PLUGIN_ROOTskill paths verified across all pluginsclauderef (Python code comment in.pyfile).mdfiles have text replacements appliedyaml.safe_load🤖 Generated with Claude Code