tiennm99/ccs
1
0
Fork 0
mirror of https://github.com/tiennm99/ccs.git synced 2026-09-05 10:18:47 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(ci): make PR-Agent reviews more verbose

Browse Source
This commit is contained in:
Tam Nhu Tran
2026-04-14 17:59:01 -04:00
parent 59c9391736
commit 4c63f1a0db
2 changed files with 21 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.pr_agent.toml
+12 -8
View File
@@ -10,23 +10,27 @@ large_patch_policy = "clip"
temperature = 0.1 temperature = 0.1
[pr_reviewer] [pr_reviewer]
require_score_review = true
require_tests_review = true require_tests_review = true
require_estimate_effort_to_review = false require_estimate_effort_to_review = true
require_can_be_split_review = true
require_security_review = true require_security_review = true
require_ticket_analysis_review = false require_todo_scan = true
require_ticket_analysis_review = true
publish_output_no_suggestions = true publish_output_no_suggestions = true
persistent_comment = true persistent_comment = true
num_max_findings = 6 num_max_findings = 10
final_update_message = false final_update_message = true
enable_review_labels_security = false enable_review_labels_security = false
enable_review_labels_effort = false enable_review_labels_effort = false
enable_intro_text = false enable_intro_text = true
enable_help_text = false enable_help_text = true
extra_instructions = """\ extra_instructions = """\
Focus on correctness, security, regressions, and missing verification. Focus on correctness, security, regressions, and missing verification.
Read the full diff before reporting findings, and inspect surrounding code before claiming a bug. Read the full diff before reporting findings, and inspect surrounding code before claiming a bug.
Prefer a short list of confirmed issues over speculative commentary. Be thorough across the enabled review sections, even when no major defects are found.
Do not pad the review with praise, scorecards, effort estimates, or generic best-practice advice. When a change looks safe, explain why it appears safe and call out any residual assumptions or manual verification gaps.
Prefer substantive analysis over generic praise or filler, and keep every claim evidence-based.
CCS-specific checks: CCS-specific checks:
- CLI output in src/ must stay ASCII-only: [OK], [!], [X], [i] - CLI output in src/ must stay ASCII-only: [OK], [!], [X], [i]
tests/unit/scripts/github/ai-review-workflow.test.ts
+9
View File
@@ -44,6 +44,15 @@ describe('PR-Agent review lane migration', () => {
expect(config).toContain('git_provider = "github"'); expect(config).toContain('git_provider = "github"');
expect(config).toContain('fallback_models = ["gpt-5.4-mini"]'); expect(config).toContain('fallback_models = ["gpt-5.4-mini"]');
expect(config).toContain('custom_model_max_tokens = 131072'); expect(config).toContain('custom_model_max_tokens = 131072');
expect(config).toContain('require_score_review = true');
expect(config).toContain('require_estimate_effort_to_review = true');
expect(config).toContain('require_can_be_split_review = true');
expect(config).toContain('require_todo_scan = true');
expect(config).toContain('require_ticket_analysis_review = true');
expect(config).toContain('num_max_findings = 10');
expect(config).toContain('final_update_message = true');
expect(config).toContain('enable_intro_text = true');
expect(config).toContain('enable_help_text = true');
expect(config).toContain('[pr_reviewer]'); expect(config).toContain('[pr_reviewer]');
expect(config).not.toContain('auto_review = true'); expect(config).not.toContain('auto_review = true');
expect(config).not.toContain('claude-code-action'); expect(config).not.toContain('claude-code-action');