feat(code-review): add classify-review-comment skill with config-driven labels - #396
Conversation
openshift-ci
Bot
added
the
do-not-merge/work-in-progress
|
Skipping CI for Draft Pull Request. |
openshift-ci
Bot
added
the
approved
|
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:
WalkthroughAdded a new Changes
Sequence Diagram(s)sequenceDiagram
participant User
participant Plugin as classify-review-comment<br/>Plugin
participant GitHub as GitHub API
participant Filter as Noise Filter
participant Classifier as Classifier
participant Config as Config
participant Output as Output
User->>Plugin: Path 1: Comment text
Plugin->>Classifier: Direct classify
Classifier->>Config: Load severity/topic labels
Classifier->>Output: severity, topic, rationale
User->>Plugin: Path 2: Comment URL
Plugin->>GitHub: Fetch single comment
GitHub-->>Plugin: Comment text
Plugin->>Classifier: Classify fetched comment
Classifier->>Config: Load labels
Classifier->>Output: severity, topic, rationale
User->>Plugin: Path 3: Classify PR comments
Plugin->>GitHub: Fetch all PR comments (paginated)
GitHub-->>Plugin: Comment list
Plugin->>Filter: Apply noise filters
Filter->>Config: Load allowed_bots, noise_patterns
Filter-->>Plugin: Filtered comments
Plugin->>Classifier: Classify each comment
Classifier->>Config: Load severity/topic labels
Classifier-->>Plugin: Per-comment results
Plugin->>Output: PR summary with aggregated counts
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes 🚥 Pre-merge checks | ✅ 8✅ Passed checks (8 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
d6105d9 to
fb341e7
Compare
bryan-cox
changed the title
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
.claude-plugin/marketplace.json (1)
187-192:⚠️ Potential issue | 🟡 MinorVersion mismatch:
code-reviewplugin version not updated in marketplace.json.The
code-reviewplugin version remains0.0.5here, butplugins/code-review/.claude-plugin/plugin.jsonwas bumped to0.0.6. This inconsistency should be fixed.🔧 Proposed fix
{ "name": "code-review", "source": "./plugins/code-review", "description": "Automated code quality review with language-aware analysis for pre-commit verification", - "version": "0.0.5" + "version": "0.0.6" },🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude-plugin/marketplace.json around lines 187 - 192, Update the version string for the entry with "name": "code-review" in marketplace.json to match the plugin's actual version (bump "version" from "0.0.5" to "0.0.6") so it stays consistent with plugins/code-review/.claude-plugin/plugin.json; ensure only the "version" value for that "code-review" object is changed and no other fields are modified.
🧹 Nitpick comments (2)
plugins/openshift/commands/api-review.md (2)
209-221: Add language specifiers to example code blocks.The example code blocks at lines 212-214 and 218-220 are missing language specifiers (flagged by markdownlint MD040).
🔧 Proposed fix
1. **Review a PR**: - ``` + ```text /openshift:api-review https://github.com/openshift/api/pull/2145 ``` Checks out the PR, runs lint and convention checks, then switches back to your branch. 2. **Review local changes**: - ``` + ```text /openshift:api-review ``` Reviews local changes against upstream master in the current openshift/api clone.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/openshift/commands/api-review.md` around lines 209 - 221, Update the two example code fences under the "Review a PR" and "Review local changes" examples to include a language specifier (use "text") by changing the opening triple-backtick markers to ```text so markdownlint MD040 is satisfied and the blocks are explicitly labeled as plain text; locate the opening fences in the "Review a PR" and "Review local changes" sections and replace them accordingly.
10-12: Add language specifier to fenced code block.Static analysis flagged this code block as missing a language specifier. Since this is a synopsis showing command usage, consider using
textorshell.🔧 Proposed fix
## Synopsis -``` +```text /openshift:api-review [pr_url]</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@plugins/openshift/commands/api-review.mdaround lines 10 - 12, The fenced
code block showing the command/openshift:api-review [pr_url]lacks a language
specifier; update the markdown fence to include a language (e.g., use ```text orthe opening triple backticks to include the chosen language token).
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@plugins/openshift/commands/api-review.md`:
- Around line 154-165: Remove the stray "thinking" code block in the
api-review.md doc (the fenced block that begins with ```thinking```), since it's
an AI reasoning placeholder; either delete it entirely or replace it with a
concise user-facing note or an HTML comment describing the validation intent
(e.g., that the doc ensures struct fields have comments, optional fields explain
omission behavior, and validation annotations are documented). Ensure the
replacement is plain prose or an HTML comment so readers aren't exposed to
implementation reasoning.
---
Outside diff comments:
In @.claude-plugin/marketplace.json:
- Around line 187-192: Update the version string for the entry with "name":
"code-review" in marketplace.json to match the plugin's actual version (bump
"version" from "0.0.5" to "0.0.6") so it stays consistent with
plugins/code-review/.claude-plugin/plugin.json; ensure only the "version" value
for that "code-review" object is changed and no other fields are modified.
---
Nitpick comments:
In `@plugins/openshift/commands/api-review.md`:
- Around line 209-221: Update the two example code fences under the "Review a
PR" and "Review local changes" examples to include a language specifier (use
"text") by changing the opening triple-backtick markers to ```text so
markdownlint MD040 is satisfied and the blocks are explicitly labeled as plain
text; locate the opening fences in the "Review a PR" and "Review local changes"
sections and replace them accordingly.
- Around line 10-12: The fenced code block showing the command
`/openshift:api-review [pr_url]` lacks a language specifier; update the markdown
fence to include a language (e.g., use ```text or ```shell) so the block becomes
a properly annotated fenced code block (change the opening triple backticks to
include the chosen language token).
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: f70952db-d8ea-46b6-b417-aaa4c38e6251
📒 Files selected for processing (7)
.claude-plugin/marketplace.jsonPLUGINS.mddocs/data.jsonplugins/code-review/.claude-plugin/plugin.jsonplugins/code-review/skills/classify-review-comment/SKILL.mdplugins/openshift/.claude-plugin/plugin.jsonplugins/openshift/commands/api-review.md
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
plugins/code-review/skills/classify-review-comment/SKILL.md (1)
160-160: Use standard phrasing in example text.Consider replacing “needs fixed” with “needs to be fixed” for clarity in the canonical example set.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/code-review/skills/classify-review-comment/SKILL.md` at line 160, Update the canonical example text in SKILL.md by replacing the phrase "needs fixed" with the standard phrasing "needs to be fixed" in the example string "hypershift-jira-solve-ci - this still needs fixed since the code did not get pushed" so the example uses clear, grammatically correct wording.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@plugins/code-review/skills/classify-review-comment/SKILL.md`:
- Around line 16-18: The fenced code block containing "labels.json (in the same
directory as this skill)" is missing a language tag which trips markdown
linters; update the backtick fence so it includes a language identifier (e.g.,
change ``` to ```text or ```json) around the block that contains labels.json to
satisfy linting and clarify the content.
- Around line 67-74: The bot-filter rules are contradictory: the rule "filter
any `*[bot]` except `coderabbitai[bot]`" conflicts with the explicit allowance
of `hypershift-jira-solve-ci[bot]`; update the policy in SKILL.md to use an
explicit allowlist/denylist or ordered precedence so behavior is
deterministic—replace the blanket "any `*[bot]` except `coderabbitai[bot]`" with
a clear allowlist that includes `coderabbitai[bot]` and
`hypershift-jira-solve-ci[bot]` (or document that
`hypershift-jira-solve-ci[bot]` is an exception) and remove the ambiguous
wildcard exception to ensure consistent classification logic referenced in the
CI bot notifications and Do classify sections.
---
Nitpick comments:
In `@plugins/code-review/skills/classify-review-comment/SKILL.md`:
- Line 160: Update the canonical example text in SKILL.md by replacing the
phrase "needs fixed" with the standard phrasing "needs to be fixed" in the
example string "hypershift-jira-solve-ci - this still needs fixed since the code
did not get pushed" so the example uses clear, grammatically correct wording.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 7fee93dd-d9d5-41f3-bd3b-51a8ed07eaf4
📒 Files selected for processing (4)
.claude-plugin/marketplace.jsonplugins/code-review/.claude-plugin/plugin.jsonplugins/code-review/skills/classify-review-comment/SKILL.mdplugins/code-review/skills/classify-review-comment/labels.json
✅ Files skipped from review due to trivial changes (2)
- .claude-plugin/marketplace.json
- plugins/code-review/skills/classify-review-comment/labels.json
🚧 Files skipped from review as they are similar to previous changes (1)
- plugins/code-review/.claude-plugin/plugin.json
|
|
||
| **Before classifying, filter out noise comments** (these carry no review signal): | ||
| - Pure slash commands: body starts with `/` followed by a command word (e.g., `/lgtm`, `/test e2e-aws`, `/approve`, `/retest`, `/cc`) | ||
| - CI bot notifications: authors like `openshift-ci-robot`, `openshift-ci[bot]`, `cwbotbot`, or any `*[bot]` author except `coderabbitai[bot]` |
There was a problem hiding this comment.
Let's make this deterministic and check against a list of approved bots. coderabbit should be allowed as the only default
|
|
||
| **Do classify** comments from: | ||
| - Human reviewers (all comments, including those directing bots) | ||
| - `coderabbitai[bot]` (substantive review comments — code issues, suggestions, questions) |
There was a problem hiding this comment.
again, the bot list should used here
There was a problem hiding this comment.
♻️ Duplicate comments (1)
plugins/code-review/skills/classify-review-comment/SKILL.md (1)
149-160:⚠️ Potential issue | 🟠 MajorExamples currently conflict with allowlist-based bot filtering.
At Line 149 and Line 154/159, these examples look like comments authored by
hypershift-jira-solve-ci, but your rule at Line 67 filters any bot not inallowed_bots(and currentlabels.jsononly allowlistscoderabbitai[bot]). That makes these examples contradictory to runtime behavior.Proposed doc fix
-**Comment:** "hypershift-jira-solve-ci - the unit test job is failing and needs fixed" +**Comment:** "@hypershift-jira-solve-ci the unit test job is failing and needs to be fixed" ```json {"severity": "required_change", "topic": "test_gap", "rationale": "Unit tests are failing — the bot made code changes without ensuring tests pass"}-Comment: "hypershift-jira-solve-ci - rebase the PR to fix the konflux issues"
+Comment: "@hypershift-jira-solve-ci rebase the PR to fix the konflux issues"{"severity": "suggestion", "topic": "ci", "rationale": "PR needs rebasing to resolve CI pipeline issues — classify by the problem (CI), not the recipient (bot)"}-Comment: "hypershift-jira-solve-ci - this still needs fixed since the code did not get pushed"
+Comment: "@hypershift-jira-solve-ci this still needs to be fixed since the code did not get pushed"{"severity": "required_change", "topic": "process", "rationale": "Code changes were not committed/pushed — a process failure, not a code issue"}</details> <details> <summary>🤖 Prompt for AI Agents</summary>Verify each finding against the current code and only fix it if needed.
In
@plugins/code-review/skills/classify-review-comment/SKILL.mdaround lines 149
- 160, The examples at Lines 149 and 154/159 in SKILL.md conflict with the
allowlist rule (see the bot filter referenced as allowed_bots near the rule at
Line 67 and labels.json allowlist), so update the example comment strings to use
an@-mentionand clearer wording (e.g., change "hypershift-jira-solve-ci -
rebase the PR..." to "@hypershift-jira-solve-ci rebase the PR..." and
"hypershift-jira-solve-ci - this still needs fixed..." to
"@hypershift-jira-solve-ci this still needs to be fixed..."), and adjust the
last example's classification JSON to reflect topic "process" and severity
"required_change" as shown in the proposed doc fix; target the examples block
around the Comment markers at Lines 149/154/159 and the JSON snippets that
follow.</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.
Duplicate comments:
In@plugins/code-review/skills/classify-review-comment/SKILL.md:
- Around line 149-160: The examples at Lines 149 and 154/159 in SKILL.md
conflict with the allowlist rule (see the bot filter referenced as allowed_bots
near the rule at Line 67 and labels.json allowlist), so update the example
comment strings to use an@-mentionand clearer wording (e.g., change
"hypershift-jira-solve-ci - rebase the PR..." to "@hypershift-jira-solve-ci
rebase the PR..." and "hypershift-jira-solve-ci - this still needs fixed..." to
"@hypershift-jira-solve-ci this still needs to be fixed..."), and adjust the
last example's classification JSON to reflect topic "process" and severity
"required_change" as shown in the proposed doc fix; target the examples block
around the Comment markers at Lines 149/154/159 and the JSON snippets that
follow.</details> --- <details> <summary>ℹ️ Review info</summary> <details> <summary>⚙️ Run configuration</summary> **Configuration used**: Path: .coderabbit.yaml **Review profile**: CHILL **Plan**: Pro **Run ID**: `9533cc07-5a19-47ea-ae5a-4abea71f5914` </details> <details> <summary>📥 Commits</summary> Reviewing files that changed from the base of the PR and between 225916515dc28e8da5abb804ef9e88cc6b0c3955 and 672f5efc81259db9b21be9517225bcdd24254c1b. </details> <details> <summary>📒 Files selected for processing (2)</summary> * `plugins/code-review/skills/classify-review-comment/SKILL.md` * `plugins/code-review/skills/classify-review-comment/labels.json` </details> <details> <summary>✅ Files skipped from review due to trivial changes (1)</summary> * plugins/code-review/skills/classify-review-comment/labels.json </details> </details> <!-- This is an auto-generated comment by CodeRabbit for review status -->
Add a skill for classifying GitHub PR review comments by severity and topic. Supports three input modes: single comment text, GitHub comment URL, or full PR URL with batch classification. Severity categories: nitpick, suggestion, required_change, question Topic categories: style, logic_bug, test_gap, api_design, documentation, ci, bot_instruction, approval, process Includes noise filtering guidance and real-world examples from openshift/hypershift PRs. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Move severity and topic label definitions from SKILL.md prose into a structured labels.json file. The skill now instructs the AI to read labels.json first and select only from the defined values, making classification more deterministic and consistent. Also removes bot_instruction topic, updates noise filters for CodeRabbit walkthroughs/review-skipped, and bumps version to 0.0.7. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
… bot filtering Address PR review comments: replace hardcoded bot names with references to the allowed_bots and noise_patterns lists in labels.json. Add text language tag to fenced code block. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
File now contains allowed_bots, noise_patterns, and classification labels — config.json better describes its broader scope. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
plugins/code-review/skills/classify-review-comment/SKILL.md (1)
82-82: Tighten wording for clarity.“subject matter” is wordy here; “subject” reads cleaner without changing meaning.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@plugins/code-review/skills/classify-review-comment/SKILL.md` at line 82, Update the wording in the list item that begins "4. **Select exactly one topic** — match the comment's subject matter to the topic descriptions and examples" by replacing "subject matter" with "subject" so the sentence reads "4. **Select exactly one topic** — match the comment's subject to the topic descriptions and examples"; locate the exact line containing that text (the list item starting with "4. **Select exactly one topic**") and make the single-word replacement to tighten the phrasing.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@plugins/code-review/skills/classify-review-comment/config.json`:
- Around line 41-87: The "topic" taxonomy array in config.json is missing the
"bot_instruction" entry referenced by the PR contract; add a new object to the
"topic" array with value "bot_instruction", a short description (e.g.,
"Instructions or commands intended for bots/automation"), and representative
examples (e.g., "run codegen", "apply auto-fix", "trigger bot task") so comments
labeled as bot directives map correctly; update the "topic" list alongside
existing entries (refer to the "topic" array and other "value" objects) to
ensure the declared contract and config stay in sync.
In `@plugins/code-review/skills/classify-review-comment/SKILL.md`:
- Around line 37-40: Documentation omits PR review-body comments (URL pattern
`#pullrequestreview-{id}`) and the corresponding GitHub API endpoints; update
the supported URL formats to include
`https://github.com/{owner}/{repo}/pull/{number}#pullrequestreview-{id}` and add
examples showing the review-body fetch endpoints `GET
/repos/{owner}/{repo}/pulls/{number}/reviews/{id}` (single) and `GET
/repos/{owner}/{repo}/pulls/{number}/reviews` (list) with jq selection for
non-empty `.body`, matching the existing issue and inline comment examples so PR
review summaries (approvals/rejections/comments) are covered.
---
Nitpick comments:
In `@plugins/code-review/skills/classify-review-comment/SKILL.md`:
- Line 82: Update the wording in the list item that begins "4. **Select exactly
one topic** — match the comment's subject matter to the topic descriptions and
examples" by replacing "subject matter" with "subject" so the sentence reads "4.
**Select exactly one topic** — match the comment's subject to the topic
descriptions and examples"; locate the exact line containing that text (the list
item starting with "4. **Select exactly one topic**") and make the single-word
replacement to tighten the phrasing.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 811ba332-d7d4-47d9-aec8-a3c266804dfb
📒 Files selected for processing (2)
plugins/code-review/skills/classify-review-comment/SKILL.mdplugins/code-review/skills/classify-review-comment/config.json
edbd932 to
264b8e0
Compare
bryan-cox
changed the title
Add #pullrequestreview-{id} URL format and /pulls/reviews API
endpoint for fetching review summaries (approvals, rejections,
general comments with state).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
openshift-ci
Bot
removed
the
do-not-merge/work-in-progress
Governance Framework ProposalI wanted to share a governance framework I've developed by working through ideas collaboratively with AI. This is meant as a reference for discussion on how we might approach AI-assisted PR review classification responsibly - not as something prescriptive. AI PR Review Classification Governance FrameworkVersion: Final v1.0 1. Executive SummaryThis framework defines a production-ready governance model for AI-assisted PR review classification. The core principle is:
AI should not replace human engineering judgment at the beginning. The final objective is graduated autonomy, not blind automation. 2. Governance Principles2.1 Human-in-the-Loop by DefaultAll AI-generated PR classifications must initially require human confirmation. Example: ai_suggestion:
label: refactor
confidence: 0.91
rationale: "code structure improvement without behavior change"Reviewer action:
2.2 Permanent Human OverrideHuman intervention rights must never be removed. human_override: true
audit_log: true
rollback_supported: trueThis is a non-negotiable engineering safety principle. 2.3 Risk-Based DelegationAutonomy must be granted by risk level, not average score only. 3. Progressive Autonomy ModelLevel 0 — Human Only (Cold Start)AI provides recommendation only. Use this phase to collect correction data. Level 1 — Human Confirm (Recommended Current Phase)AI proposes labels. Human must confirm every result. This is the best stage for current rollout. Level 2 — Conditional AutonomyAI can automatically classify low-risk labels only. Example: auto_apply:
- nit
- docs
- style
- typo
- naming
manual_required:
- blocker
- must-fix
- security
- performance
- breaking-changeThis is the recommended delegation model. Level 3 — AI Primary + Human OverrideAI becomes primary classifier. Human remains escalation and override authority. 4. Confidence Gating ModelConfidence score must control workflow routing. confidence_thresholds:
auto_apply: 0.95
human_review: 0.80
manual_required: 0.00Routing logic: Confidence >= 0.95Auto apply allowed for low-risk labels only. 0.80 <= Confidence < 0.95Human confirmation required. Confidence < 0.80Mandatory manual classification. 5. Human Scoring FrameworkDo NOT use a simple 1–10 score only. Use structured evaluation. review_score:
correctness: 5
severity_assessment: 5
rationale_quality: 4
confidence_calibration: 4Weighted formula: Delegation threshold recommendation: Only then move to higher autonomy. 6. Experience Learning FrameworkCritical requirement:
Bad example: ai_label: enhancement
human_label: blockerGood example: misclassification_case:
ai_label: enhancement
human_label: blocker
root_cause: "failed to recognize upgrade-path regression risk"
lesson: "compatibility-related comments should be elevated"This is the true experience accumulation layer. 7. Feedback Loop ArchitectureThis should become a continuous iteration engine. 8. Recommended Production PolicyRecommended immediate rollout: phase: Level 1
auto_apply:
- docs
- nit
- style
manual_required:
- security
- architecture
- blocker
- performance9. Final RecommendationThe correct strategy is:
This is the safest and most enterprise-ready AI governance framework for PR review classification. This framework aligns with the current implementation in this PR, where the skill outputs classifications but doesn't automatically apply them (Level 1). Happy to discuss or revise based on feedback! |
…omment Add a weighted signal rubric (0.00–1.00) for classification confidence, modeled after the payload analysis scoring approach. Signals include signal word match, unambiguous category, example pattern match, context reinforcement, and single viable label. Confidence thresholds gate workflow routing: >= 0.95 auto-apply, 0.80–0.94 human review, < 0.80 manual classification. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
@wangke19 there should be automatic confidence scores applied now. |
| "examples": ["add a test for empty input", "unit tests failing", "test doesn't validate X"] | ||
| }, | ||
| { | ||
| "value": "api_design", |
There was a problem hiding this comment.
can we include architecture design and security?
There was a problem hiding this comment.
Added architecture_design and security as new topic categories in config.json and SKILL.md with descriptions, signal examples, and real-world classification examples.
…opics Add two new topic categories per review feedback: - architecture_design: system architecture, component boundaries, patterns - security: vulnerabilities, auth issues, input validation, secrets Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
/lgtm |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: bryan-cox, enxebre 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 |
3 participants
Summary
classify-review-commentskill for classifying GitHub PR review comments by severity and topicconfig.jsondefines all valid classification values, allowed bots, and noise patterns — making classification deterministicnitpick,suggestion,required_change,question,unclassifiedstyle,logic_bug,test_gap,api_design,documentation,ci,approval,process,unclassifiedallowed_botslist in config.json (default:coderabbitai[bot]only)noise_patternslist in config.jsonContext
The jira-agent performance dashboard tracks AI-generated PRs and their review feedback. This skill enables classifying PR comments by severity and topic to identify patterns in what reviewers catch most often.
Test plan
"small nit: rename cnt to count"https://github.com/openshift/hypershift/pull/7620allowed_bots(config.json) are classified🤖 Generated with Claude Code