.pr_agent.toml

[config]
model = "openai/minimax-M3"
fallback_models = []
custom_model_max_tokens = 16384

[pr_reviewer]
# Skip ticket compliance for release-please PRs (they only update CHANGELOG)
ignore_pr_title = ["^\\[Auto\\]", "^Auto", "^chore\\(main\\): release"]
# Require tests for non-docs PRs
require_tests_review = true
# Enable security review
require_security_review = true
# Enable effort estimation
require_estimate_effort_to_review = true
# Enable ticket analysis (but skip for release PRs)
require_ticket_analysis_review = true
# Max findings per review
num_max_findings = 5
# Publish even if no suggestions
publish_output_no_suggestions = true
# Use persistent comment (update instead of new comment)
persistent_comment = true
# Enable intro text with summary
enable_intro_text = true
# Disable help text (reduces noise)
enable_help_text = false
# Enable coverage footer
enable_review_coverage_footer = true
# Extra instructions for better reviews
extra_instructions = """
Focus on:
1. Code correctness and potential bugs
2. Security vulnerabilities
3. Test coverage gaps
4. Scope creep (changes unrelated to PR title/description)
5. API compatibility breaks

IMPORTANT: If the PR title starts with "chore(main): release" or the PR description says "this pr was generated with release please", DO NOT perform ticket compliance analysis. Release PRs only update CHANGELOG.md and version numbers - there is no code to review against ticket acceptance criteria. Just note "Release PR - no code changes to review" and skip.

For docs-only PRs:
- Check for factual accuracy
- Verify code examples are correct
- Check for broken links
"""