Repository navigation
test(OMN-10400): prove orchestrated merge sweep and session baseline - #465
Merged
Merged
Conversation
|
Warning Rate limit exceeded
To keep reviews running without waiting, you can enable usage-based add-on for your organization. This allows additional reviews beyond the hourly cap. Account admins can enable it under billing. ⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (40)
📝 WalkthroughWalkthroughAdds a CLI reporting system with typed Pydantic models and multiple renderers, a new operator script to run the orchestrated merge-sweep workflow (with live effects and git worktree handling), tri-state PR required-check evaluation, PR-polish command emission, deterministic session orchestration fixtures, and broad test coverage and contract metadata. Changes
Sequence Diagram(s)sequenceDiagram
participant Caller
participant Script as run_orchestrated_merge_sweep_workflow
participant GitHub
participant Compute as NodeMergeSweep
participant Triage as TriageOrchestrator
participant Reducer as MergeSweepStateReducer
participant Effect as EffectHandlers
participant Classify as SweepOutcomeClassify
Caller->>Script: invoke (--repos, --execute?)
Script->>GitHub: fetch branch protection & open PRs
Script->>Compute: compute merge-sweep (ModelMergeSweepRequest)
Compute-->>Script: ModelMergeSweepResult
Script->>Triage: triage (emit commands)
Triage-->>Script: command list (merge/ci/polish/etc)
Script->>Reducer: reduce commands -> state
Reducer-->>Script: ModelMergeSweepState
alt --execute
loop for each command
Script->>Effect: execute effect async (auto-merge/ci-rerun/polish)
Effect-->>Script: dispatch event
Script->>Classify: classify event -> outcome
Classify-->>Script: outcome enum
end
else dry-run
Script->>Script: log planned commands
end
Script->>Script: summarize counts & exit with code
Script-->>Caller: exit code (0/1)
sequenceDiagram
participant Triage as TriageOrchestrator
participant Request as ModelTriageRequest
participant TrackB as Track B PR
participant Polish as PRPolishHandler
participant Classify as SweepOutcomeClassify
Triage->>Request: inspect emit_pr_polish_commands & dry_run
alt emit_pr_polish_commands == true
Triage->>TrackB: evaluate required checks
alt checks failed (terminal)
Triage->>Polish: emit ModelPrPolishStartCommand (dry_run flag)
Polish-->>Triage: ModelPrPolishCompletedEvent (extra.final_phase)
Triage->>Classify: classify PR_POLISH_COMPLETED (final_phase)
Classify-->>Triage: SUCCESS/FAILED/DEGRADED
else checks not failed
Triage->>Triage: skip polish emission
end
else
Triage->>Triage: suppress polish emission
end
Triage->>Triage: deduplicate total_prs by (repo, pr_number)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~50 minutes Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Review rate limit: 0/5 reviews remaining, refill in 3 minutes and 8 seconds. Comment |
jonahgabriel
enabled auto-merge
April 30, 2026 15:45
|
| Verdict | Meaning | Blocks merge? |
|---|---|---|
passed |
No critical/error findings | No |
blocked |
CRITICAL or ERROR findings found | Yes |
degraded |
All models unavailable (infra) | No (pilot) |
Powered by omniintelligence.review_pairing.cli_review — node-based adversarial review via HandlerLlmCliSubprocess (OMN-8524)
jonahgabriel
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: 10
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
src/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.py (1)
355-383:⚠️ Potential issue | 🟠 Major | ⚡ Quick winGate reruns and rebases on terminal failures, not any non-pass state.
These branches still use
not pr.required_checks_pass, so aB_POLISHpayload withrequired_checks_pending=Truecan fan out a live CI rerun/rebase. The compute side now distinguishes terminal failure from pending; this orchestrator needs the same predicate or the tri-state fix is easy to bypass with older or manually-constructed inputs.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.py` around lines 355 - 383, The current branches for CI rerun/rebase use the brittle predicate "not pr.required_checks_pass" which treats pending checks as failures; update both conditional guards that now read "and not pr.required_checks_pass" (the MERGEABLE+BLOCKED rerun path that calls _resolve_failing_run_id and returns ModelCiRerunCommand, and the MERGEABLE+BEHIND rebase path) to explicitly require terminal failure instead of any non-pass state—i.e., replace the predicate with a tri-state-aware check such as "and pr.required_checks_failed" or "and (not pr.required_checks_pass and not pr.required_checks_pending)" depending on which property exists in the PR model so pending checks do not trigger reruns/rebases.src/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.py (1)
465-489:⚠️ Potential issue | 🟠 Major | ⚡ Quick winUse the reducer’s emitted merge order here.
triage_recordsis created fromtrack_abefore the orchestrator runs, and lines 479-483 rebuildorderedfrom that same tuple. As written,use_lifecycle_ordering=Truejust round-trips the original Track A order instead of applying the lifecycle/reducer ordering.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.py` around lines 465 - 489, The current code rebuilds ordered from triage_records (which mirrors track_a) so use_lifecycle_ordering does not apply the reducer/orchestrator merge ordering; change the construction to iterate the reducer-emitted merge sequence (the merge_prs passed by the orchestrator / recorded by _NoopMerge) instead of triage_records: keep the pr_lookup = {(c.pr.repo, c.pr.number): c for c in track_a} and the seen set, then for each entry in merge_prs (or whatever variable holds the reducer's MERGE intent order) form the same key and append pr_lookup[key] when present, then finally append any remaining track_a stragglers as before; ensure you reference the reducer/orchestrator merge_prs variable rather than triage_records so use_lifecycle_ordering actually applies the lifecycle order.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@docs/reference/2026-04-28-market-skill-baseline.md`:
- Around line 3-5: Remove the regenerated, timestamped execution transcript and
any volatile capture state from the public baseline doc (e.g., strings like
`2026-04-30T16:49:31.241546+00:00`, `Baseline window: 2026-04-30`, and the `Repo
root: <omnimarket>` note); retain only stable, contract-level baseline facts and
descriptions. Move the detailed historical run output and any workspace/ticket
identifiers to a non-public/generated artifact or internal storage, and replace
the removed sections with a short note pointing to the internal artifact or
stating that transcripts are available privately.
In `@scripts/run_orchestrated_merge_sweep_workflow.py`:
- Around line 229-263: _prepare_polish_worktree currently always uses the global
REPO_ROOT and runs git operations there which breaks when repo != the local
checkout; change it to resolve the target checkout for the incoming repo (i.e.
call the existing repo→checkout resolver or add a new helper) and use that
checkout path in place of REPO_ROOT for the _pr_head call,
current_sha/_is_clean_worktree checks, _worktrees_for_branch lookup, and the git
_run fetch/worktree commands, or if you do not yet support multi-repo workspaces
add a hard guard at the top of _prepare_polish_worktree that raises if repo is
not the local repo name (e.g. not "omnimarket"); update references to REPO_ROOT
in this function to the resolved checkout Path so all branch/SHA checks and git
fetch/worktree operations act against the correct repository.
- Around line 371-406: The ModelPrPolishStartCommand branch currently prepares a
worktree and calls run_live_pr_polish even when execute is false; guard that
branch with the same execute short-circuit used below so planning/dry-run
doesn't perform git actions: before calling _prepare_polish_worktree, check the
execute flag (the same condition used by the existing "if not execute" block)
and if not executing just _log the planned command (use _log and _dump_model on
command) and return []; otherwise proceed to call _prepare_polish_worktree,
run_live_pr_polish, and always call _cleanup_polish_worktree in the finally
block as currently implemented.
In `@src/omnimarket/cli/reporting.py`:
- Around line 137-177: The report builder build_report_from_merge_sweep_result
currently hardcodes status = EnumMarketCliStatus.SUCCESS; change it to derive
the report status from result.status (e.g., map result.status == "error" to
EnumMarketCliStatus.FAILURE and result.status == "warning"/"partial" to an
appropriate enum value, otherwise SUCCESS) and use that mapped value when
constructing ModelMarketCliReport so the report status matches the actual
result.status.
In `@src/omnimarket/nodes/node_merge_sweep_compute/__main__.py`:
- Around line 101-112: The check-evaluation logic currently only treats
conclusion == "SUCCESS" as passing and misses legacy StatusContext success and
non-failing CheckRun conclusions; update the block that computes conclusion and
status (variables conclusion, status, and the surrounding loop that sets
failed/pending using failed_conclusions) so that it treats status == "SUCCESS"
as a passing check and treats conclusion in {"SUCCESS","NEUTRAL","SKIPPED"} as
passing as well (i.e., skip marking failed or pending when conclusion is one of
those), leaving failed_conclusions handling intact for explicit failure reasons.
In
`@src/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.py`:
- Around line 545-548: The code only treats checks with conclusion == "FAILURE"
as failures when extracting details_url/run_id, but
node_merge_sweep_compute.consumer._required_check_state() treats
ACTION_REQUIRED, CANCELLED, FAILED, STARTUP_FAILURE, STALE, and TIMED_OUT as
terminal failures too; update the condition in the resolver(s) (the blocks using
check.get("conclusion"), details_url, and _run_id_from_details_url, including
the other resolver at the other occurrence) to test membership in a failure set
(e.g.
{"FAILURE","ACTION_REQUIRED","CANCELLED","FAILED","STARTUP_FAILURE","STALE","TIMED_OUT"})
so run/job resolution and failing-job-name scanning run for all terminal failure
conclusions.
In
`@src/omnimarket/nodes/node_session_orchestrator/handlers/handler_session_orchestrator.py`:
- Around line 1434-1448: The handler currently mutates process-wide os.environ
keys ONEX_STATE_DIR and ONEX_PARENT_SESSION_ID around calling
NodeDispatchWorker().handle(command), causing non-reentrancy and races; instead,
change HandlerSessionOrchestrator to stop touching os.environ and pass the
context explicitly into NodeDispatchWorker (e.g., construct
NodeDispatchWorker(state_dir=state_dir, parent_session_id=session_id) or call
NodeDispatchWorker().handle(command, state_dir=state_dir,
parent_session_id=session_id)), and update the NodeDispatchWorker class
(constructor and/or handle method) to accept and use those parameters internally
wherever it currently reads os.environ for
ONEX_STATE_DIR/ONEX_PARENT_SESSION_ID. Ensure no global env mutation remains in
handler_session_orchestrator while preserving existing behavior by using the
passed-in values.
- Around line 1050-1055: The code currently treats an empty result from
self._load_linear_fixture when ONEX_SESSION_ORCHESTRATOR_LINEAR_FIXTURE is set
as success by returning [], which silences setup failures; change the behavior
in the handler where fixture_path =
os.environ.get("ONEX_SESSION_ORCHESTRATOR_LINEAR_FIXTURE", "") is read so that
if fixture_path is set but self._load_linear_fixture(fixture_path) returns an
empty or falsy tickets list, the handler fails fast (raise a clear exception or
return an error status) instead of returning []—update the block around
fixture_path/_load_linear_fixture in handler_session_orchestrator.py to log an
explicit error including fixture_path and either raise a ValueError or propagate
an error state so Phase 2 does not complete successfully when the configured
fixture yields zero tickets.
- Around line 1142-1162: The fixture loader _load_linear_fixture currently only
validates the top-level container and can return nodes that break _score_tickets
which expects GraphQL-style nested objects (e.g., labels["nodes"],
children["nodes"]); update _load_linear_fixture to validate each node's shape
before returning: ensure each node is a dict and that required nested fields
like "labels" and "children" exist and are dicts with a "nodes" list (or coerce
simple lists into {"nodes": [...]}) and otherwise raise SessionLinearFetchError
or skip the malformed node so that _score_tickets never receives invalid shapes;
reference the function names _load_linear_fixture and _score_tickets when making
the change.
In `@tests/unit/nodes/node_ticket_pipeline/test_main_cli_outputs.py`:
- Around line 23-35: The subprocess.run call in
tests/unit/nodes/node_ticket_pipeline/test_main_cli_outputs.py should include a
timeout to avoid CI hangs; add a timeout parameter (e.g. timeout=30) to the
subprocess.run invocation that runs omnimarket.nodes.node_ticket_pipeline and
wrap the call in a try/except that catches subprocess.TimeoutExpired then fail
the test explicitly (raise or assert False with a clear message) so a hung CLI
produces a deterministic test failure rather than stalling CI.
---
Outside diff comments:
In
`@src/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.py`:
- Around line 465-489: The current code rebuilds ordered from triage_records
(which mirrors track_a) so use_lifecycle_ordering does not apply the
reducer/orchestrator merge ordering; change the construction to iterate the
reducer-emitted merge sequence (the merge_prs passed by the orchestrator /
recorded by _NoopMerge) instead of triage_records: keep the pr_lookup =
{(c.pr.repo, c.pr.number): c for c in track_a} and the seen set, then for each
entry in merge_prs (or whatever variable holds the reducer's MERGE intent order)
form the same key and append pr_lookup[key] when present, then finally append
any remaining track_a stragglers as before; ensure you reference the
reducer/orchestrator merge_prs variable rather than triage_records so
use_lifecycle_ordering actually applies the lifecycle order.
In
`@src/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.py`:
- Around line 355-383: The current branches for CI rerun/rebase use the brittle
predicate "not pr.required_checks_pass" which treats pending checks as failures;
update both conditional guards that now read "and not pr.required_checks_pass"
(the MERGEABLE+BLOCKED rerun path that calls _resolve_failing_run_id and returns
ModelCiRerunCommand, and the MERGEABLE+BEHIND rebase path) to explicitly require
terminal failure instead of any non-pass state—i.e., replace the predicate with
a tri-state-aware check such as "and pr.required_checks_failed" or "and (not
pr.required_checks_pass and not pr.required_checks_pending)" depending on which
property exists in the PR model so pending checks do not trigger reruns/rebases.
🪄 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: defaults
Review profile: CHILL
Plan: Pro
Run ID: 938077f1-91f4-46d1-a2ff-7e50c0cda443
📒 Files selected for processing (54)
contracts/OMN-10400.yamldocs/reference/2026-04-28-market-skill-baseline.jsondocs/reference/2026-04-28-market-skill-baseline.mdscripts/run_orchestrated_merge_sweep_workflow.pysrc/omnimarket/cli/__init__.pysrc/omnimarket/cli/args.pysrc/omnimarket/cli/output/__init__.pysrc/omnimarket/cli/output/handlers/__init__.pysrc/omnimarket/cli/output/handlers/handler_json.pysrc/omnimarket/cli/output/handlers/handler_markdown.pysrc/omnimarket/cli/output/handlers/handler_text.pysrc/omnimarket/cli/output/handlers/handler_yaml.pysrc/omnimarket/cli/output/protocol.pysrc/omnimarket/cli/output/registry.pysrc/omnimarket/cli/reporting.pysrc/omnimarket/market_skill_baseline.pysrc/omnimarket/models/cli_report.pysrc/omnimarket/nodes/node_merge_sweep_compute/__main__.pysrc/omnimarket/nodes/node_merge_sweep_compute/consumer.pysrc/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.pysrc/omnimarket/nodes/node_merge_sweep_triage_orchestrator/contract.yamlsrc/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.pysrc/omnimarket/nodes/node_merge_sweep_triage_orchestrator/models/model_triage_request.pysrc/omnimarket/nodes/node_session_orchestrator/handlers/handler_session_orchestrator.pysrc/omnimarket/nodes/node_session_orchestrator/tests/test_handler_session_orchestrator.pysrc/omnimarket/nodes/node_sweep_outcome_classify/handlers/handler_outcome_classify.pysrc/omnimarket/nodes/node_sweep_outcome_classify/models/model_sweep_outcome.pysrc/omnimarket/nodes/node_ticket_pipeline/__main__.pytests/test_codex_runtime_client.pytests/test_golden_chain_merge_sweep.pytests/test_golden_chain_merge_sweep_executor.pytests/test_golden_chain_ticket_pipeline.pytests/test_market_skill_smokes.pytests/test_no_cross_node_reach_in.pytests/test_orchestrated_merge_sweep_workflow.pytests/test_sweep_outcome_classify.pytests/test_triage_orchestrator.pytests/test_triage_phase2_emit_rules.pytests/unit/cli/__init__.pytests/unit/cli/_fixtures.pytests/unit/cli/output/__init__.pytests/unit/cli/output/handlers/__init__.pytests/unit/cli/output/handlers/test_handler_json.pytests/unit/cli/output/handlers/test_handler_markdown.pytests/unit/cli/output/handlers/test_handler_text.pytests/unit/cli/output/handlers/test_handler_yaml.pytests/unit/cli/output/test_registry.pytests/unit/cli/test_args.pytests/unit/cli/test_reporting.pytests/unit/models/test_cli_report.pytests/unit/nodes/node_merge_sweep_compute/__init__.pytests/unit/nodes/node_merge_sweep_compute/test_main_cli_outputs.pytests/unit/nodes/node_ticket_pipeline/__init__.pytests/unit/nodes/node_ticket_pipeline/test_main_cli_outputs.py
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 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/run_orchestrated_merge_sweep_workflow.py`:
- Around line 409-417: Polish evidence is being returned as a reducer outcome
which blocks later real remediation outcomes; stop emitting a reducible outcome
for polish by not returning the classified outcome from the polish handler.
Specifically, in run_orchestrated_merge_sweep_workflow.py replace the return of
_classify_outcome(_polish_completed_to_input(...)) (functions
_polish_completed_to_input and _classify_outcome) with an empty list (or
otherwise mark the result as non-reducer-only evidence) so
ModelPrPolishStartCommand/HandlerTriageOrchestrator polish events do not consume
the HandlerMergeSweepStateReducer dedup slot keyed by repo#pr_number.
- Around line 201-214: The function _worktrees_for_branch currently matches
branches by suffix (raw_branch.endswith(f"/{branch}")), which can incorrectly
match unrelated refs; update _worktrees_for_branch to compare the full ref
exactly (e.g., raw_branch == f"refs/heads/{branch}") so only worktrees whose
branch ref equals the exact ref are returned, and ensure any callers like
_prepare_polish_worktree use the corrected _worktrees_for_branch result.
🪄 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: defaults
Review profile: CHILL
Plan: Pro
Run ID: 7354c1d4-3183-46a4-b013-983ca3b36ed7
📒 Files selected for processing (4)
scripts/run_orchestrated_merge_sweep_workflow.pysrc/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.pytests/test_golden_chain_merge_sweep.pytests/test_orchestrated_merge_sweep_workflow.py
🚧 Files skipped from review as they are similar to previous changes (3)
- src/omnimarket/nodes/node_merge_sweep_compute/handlers/handler_merge_sweep.py
- tests/test_golden_chain_merge_sweep.py
- tests/test_orchestrated_merge_sweep_workflow.py
Contributor
Author
|
Added the next Codex skill shim: Scope:
Proof:
Commit: |
Contributor
Author
|
Added the next Codex skill shim: Scope:
Proof:
Commit: |
Contributor
Author
|
Added the next Codex skill shim: Scope:
Proof:
Commit: |
Contributor
Author
|
Added the final missing Codex skill shim in this pass: Scope:
Proof:
Commit: |
Contributor
Author
|
Corrected focus back to actual market-skill end-to-end proof. Ran the full market-owned skill baseline, not just Codex shim checks:
Result:
Committed refreshed durable evidence: Note: the current baseline runner is batch-style and prints after completion. If we want the exact operator experience of logs as each node runs inline, the next concrete improvement should be a streaming market-skill proof runner rather than more skill packaging. |
Contributor
Author
|
Added streaming market-skill proof output. New behavior:
Proof run: Result:
Additional checks:
Commit: |
Contributor
Author
|
Pushed hardening fixes in Local verification after the merge:
All previously open CodeRabbit review threads are resolved. |
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
1 participant
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.
Ticket
OMN-10400
Summary
Live proof
uv run python scripts/run_orchestrated_merge_sweep_workflow.py --repos OmniNode-ai/omnimarket --executeci_rerun=1,armed=2,failed=0,terminal_emitted=True.Runtime adapter coverage now in baseline
aislop_sweep: adapter dispatch scans a controlled repo and returns real findings.pr_lifecycle_orchestrator/merge-sweep: adapter dispatch runs the real PR lifecycle FSM with deterministic sub-handlers and writesmerge-sweep/<run_id>/result.json.pr_polish: adapter dispatch runs the pure polish FSM todone.local_review: adapter dispatch runs the pure review FSM todone.coderabbit_triage: adapter dispatch runs real CodeRabbit classification over deterministic GraphQL review-thread shapes.session_bootstrap: adapter dispatch runs the real handler, writes contract snapshot and cron IDs.session_orchestrator: adapter dispatch runs the real handler, scores fixture Linear tickets, compiles dispatch specs, and writes durable session artifacts.ticket_pipeline: adapter dispatch runs PRE_FLIGHT plus compile-only IMPLEMENT and honestly stops at unwired LOCAL_REVIEW.Verification
PatternBBrokerClienttoCodexRuntimeRequestAdapter, removed the misleading public broker type names, and updated source plus packaged Codex skill shims to describe the handler/adapter path.uv run ruff format --check src/omnimarket/adapters/codex/runtime_client.py tests/test_codex_runtime_client.py tests/test_codex_plugin_shims.py scripts/generate_adapters.py src/omnimarket/adapters/codex/topics.py-> passeduv run ruff check src/omnimarket/adapters/codex/runtime_client.py tests/test_codex_runtime_client.py tests/test_codex_plugin_shims.py scripts/generate_adapters.py src/omnimarket/adapters/codex/topics.py-> passeduv run pytest tests/test_codex_runtime_client.py tests/test_install_codex_skills.py tests/test_codex_plugin_shims.py -q-> 37 passeduv run pytest tests/test_codex_runtime_client.py tests/test_golden_chain_ticket_pipeline.py tests/test_market_skill_smokes.py -q-> 52 passeduv run pytest tests/test_codex_runtime_client.py tests/test_golden_chain_coderabbit_triage.py tests/test_market_skill_smokes.py -q-> 61 passeduv run pytest tests/test_codex_runtime_client.py tests/unit/nodes/node_pr_lifecycle_orchestrator/test_main_cli.py tests/test_golden_chain_pr_lifecycle_orchestrator.py tests/test_market_skill_smokes.py -q-> 61 passeduv run pytest tests/test_codex_runtime_client.py src/omnimarket/nodes/node_session_orchestrator/tests/test_handler_session_orchestrator.py tests/test_market_skill_smokes.py -q-> 60 passeduv run mypy --strict src/omnimarket/nodes/node_merge_sweep_triage_orchestrator/handlers/handler_triage.py src/omnimarket/nodes/node_session_orchestrator/handlers/handler_session_orchestrator.py src/omnimarket/market_skill_baseline.pyuv run python scripts/run_market_skill_baseline.py --strict --json-out docs/reference/2026-04-28-market-skill-baseline.json --markdown-out docs/reference/2026-04-28-market-skill-baseline.md-> 8 working / 0 degraded / 0 failinguv run ruff check tests/test_codex_runtime_client.py src/omnimarket/market_skill_baseline.py && uv run ruff format --check tests/test_codex_runtime_client.py src/omnimarket/market_skill_baseline.pySummary by CodeRabbit
New Features
Bug Fixes
Documentation
Tests