Skip to content

Add manual AI review tiers#671

Draft
MauroToscano wants to merge 13 commits into
mainfrom
feat/manual-ai-review-tiers
Draft

Add manual AI review tiers#671
MauroToscano wants to merge 13 commits into
mainfrom
feat/manual-ai-review-tiers

Conversation

@MauroToscano

Copy link
Copy Markdown
Contributor

Summary

  • make existing Codex, Claude, and Kimi review workflows manual comment-triggered only
  • add /ai-review standard and /ai-review critical tiers
  • document tier policy, model/prompt strategy, OpenRouter matrix plan, and evaluation artifacts

Validation

  • git diff --check
  • parsed updated workflow YAML with Ruby YAML.load_file

Note: actionlint is not installed in this workspace.

@MauroToscano

Copy link
Copy Markdown
Contributor Author

/ai-review standard

@MauroToscano MauroToscano added the ai-review-standard Trigger the standard AI review workflow label Jun 16, 2026
@github-actions

github-actions Bot commented Jun 16, 2026

Copy link
Copy Markdown

AI Review (standard)

PR #671 · 17 changed files

Findings

No non-rejected structured findings were reported.

Reviewer Lanes

Lane Model Prompt Status Findings
minimax-correctness minimax/minimax-m3 correctness success: parse warning: invalid JSON in fenced block: Expecting ',' delimiter 0

Raw lane outputs, candidates, final issues, and model metrics are uploaded as workflow artifacts.

@MauroToscano MauroToscano added ai-review-standard Trigger the standard AI review workflow and removed ai-review-standard Trigger the standard AI review workflow labels Jun 16, 2026
@MauroToscano MauroToscano added ai-review-standard Trigger the standard AI review workflow and removed ai-review-standard Trigger the standard AI review workflow labels Jun 16, 2026
@MauroToscano MauroToscano force-pushed the feat/manual-ai-review-tiers branch from 1b27986 to 52633cb Compare June 16, 2026 19:16
@MauroToscano MauroToscano added ai-review-standard Trigger the standard AI review workflow and removed ai-review-standard Trigger the standard AI review workflow labels Jun 16, 2026
@MauroToscano MauroToscano added ai-review-standard Trigger the standard AI review workflow and removed ai-review-standard Trigger the standard AI review workflow labels Jun 16, 2026
response_format={type: json_object} was added in the hardening commit and
turned out to be the cause of empty model responses: it routes to
structured-output providers and makes reasoning models (minimax-m3, glm,
mimo) reason until truncated at max_tokens without ever emitting content
(observed reasoning_tokens=33989, completion_tokens=32000, findings=0).

Make response_format opt-in per lane and rely on the existing extract_json
parser, matching the request shape that works locally. Also capture
finish_reason in the lane result so truncation is visible in the report.
@MauroToscano MauroToscano added ai-review-standard Trigger the standard AI review workflow and removed ai-review-standard Trigger the standard AI review workflow labels Jun 16, 2026
Without forced JSON mode the model occasionally emits invalid JSON (e.g.
unescaped quotes when a finding quotes code), which strict json.loads
rejects all-or-nothing, dropping a whole review to zero findings.

Add an optional json-repair fallback in extract_json: try strict parsing
first, and only on failure fall back to repair, flagging it as a parse
warning so invalid output stays visible. Install json-repair in the
review/verifier lane steps. Verified against real lane output: recovers
all 6 findings that strict parsing dropped.
@MauroToscano MauroToscano removed the ai-review-standard Trigger the standard AI review workflow label Jun 16, 2026
@MauroToscano MauroToscano added the ai-review-standard Trigger the standard AI review workflow label Jun 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ai-review-standard Trigger the standard AI review workflow

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant