workflows: Add issue-led PR readiness gate - #11
Conversation
The existing reviewer could publish actionable findings while its Check Run still succeeded, and it did not verify that a PR followed an implementation- ready Issue plan on every pushed head. - Add deterministic and AI Issue-readiness review contracts based on write-issue - Bind PR metadata, native Issue linkage, plan conformance, code findings, policy, and workflow source into PASS/FAIL evidence - Publish a separate OpenAI PR readiness Check Run and refresh reviews on every relevant PR or Issue change - Cover snapshot invalidation, format failures, incremental review, and fail-closed behavior with focused tests and rollout documentation Generated with [Codex](https://github.com/openai)
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: 70f905afb9
Reviewed range: e16db5d052..70f905afb9
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 075411674fc7b4159f7225572b6547631629bbff55010c92df4729c3f5255a49
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 9s
Input tokens: 420,897
Cached input tokens: 366,500
Cache write tokens: 54,364
Cache hit ratio: 87.1%
Output tokens: 5,580
Reasoning tokens: 3,792
Total tokens: 426,477
Estimated Codex credits: 7.783
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:fb212634474f38b515bb197cfa4a0c0f2ace70fa9b8132384cef1e163e644ce6 |
1m 4s | 313,016 | 260,534 | 83.2% | 52,455 | 5,153 | 3,703 | 318,169 | 6.841 |
aggregate:075411674fc7b4159f7225572b6547631629bbff55010c92df4729c3f5255a49 |
5s | 107,881 | 105,966 | 98.2% | 1,909 | 427 | 89 | 108,308 | 0.942 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: FAIL
Evidence: 17585bc90a04bd2a4d8d46345f18d9343e5fa8277e4ba3a5fffc22b61e2753d4
Readiness blockers
code-review.github/workflows/codex-openai-issue-review.yml:47: Keep the write-capable Issue token out of the model-review job
One security issue found.
1 inline finding published.
The Issue reviewer processed untrusted model input in the same job that held write access to repository Issues, weakening the intended publication boundary. - Keep deterministic and model Issue review in a read-only job - Pass only structured review evidence to a dependent publication job - Grant issues write permission exclusively to the comment publication job Generated with [Codex](https://github.com/openai)
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: cda3f3e63c
Reviewed range: e16db5d052..cda3f3e63c
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 3924adfcbb20c6be2ae3c6d09c184aee3f99f26037a9a5516679672b75b89edb
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 51s
Input tokens: 547,888
Cached input tokens: 474,348
Cache write tokens: 73,507
Cache hit ratio: 86.6%
Output tokens: 5,111
Reasoning tokens: 1,633
Total tokens: 552,999
Estimated Codex credits: 9.478
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:0a9dafec7ed59868f624e6f790851dd742d7858db43cd1b5b429fdc7582fffde |
47s | 401,705 | 330,359 | 82.2% | 71,319 | 4,749 | 1,582 | 406,454 | 8.305 |
aggregate:3924adfcbb20c6be2ae3c6d09c184aee3f99f26037a9a5516679672b75b89edb |
4s | 146,183 | 143,989 | 98.5% | 2,188 | 362 | 51 | 146,545 | 1.173 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: FAIL
Evidence: 13aad7b2ac09076a52ddce99bbe77f054209df96c642a651c1fe76a6d2409e6e
Readiness blockers
code-review.github/workflows/codex-openai-issue-review.yml:82: Read the GraphQL repository object before resolving the Issue
The reusable Issue readiness workflow cannot resolve any Issue because it reads the GraphQL response at the wrong level.
1 inline finding published.
The Issue readiness workflow read the GraphQL response wrapper as though it were the repository object, causing every Issue lookup to fail before review. - Dereference the repository field before reading Issue and default-branch data - Add a source-level regression assertion for the GraphQL response shape Generated with [Codex](https://github.com/openai)
There was a problem hiding this comment.
❌ PR readiness: FAIL (2 blockers)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: 335b9a87e2
Reviewed range: e16db5d052..335b9a87e2
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 475d9df5942c9d243246c32dc2208cbfea714d1c392af0308b895a1996fbd56a
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 24s
Input tokens: 388,838
Cached input tokens: 335,694
Cache write tokens: 53,111
Cache hit ratio: 86.3%
Output tokens: 6,572
Reasoning tokens: 4,532
Total tokens: 395,410
Estimated Codex credits: 7.884
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:103d61157e766c6df8c8d044f659e8228a597bf99ae905d67a1b32096e7231e5 |
1m 19s | 283,549 | 232,981 | 82.2% | 50,541 | 6,110 | 4,468 | 289,659 | 6.908 |
aggregate:475d9df5942c9d243246c32dc2208cbfea714d1c392af0308b895a1996fbd56a |
5s | 105,289 | 102,713 | 97.6% | 2,570 | 462 | 64 | 105,751 | 0.976 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: FAIL
Code review: FAIL
Evidence: 0d6630774afc31568341a689971885017dcc8f458b83e1cb950c90fbd7dc8ed6
Readiness blockers
plan-conformanceThe linked Issue requires tests for linkage variants, stale or failed Issue evidence, plan deviations, thread handling, API failures, same-head metadata and Issue invalidation, synchronize invalidation, and full PASS. The added issue-review and pr-readiness tests cover only small deterministic fixtures and a basic code-finding aggregation; they do not exercise those required scenarios.code-review.github/workflows/codex-openai-review.yml:1373: Revalidate the complete readiness identity before publishing success
Readiness cannot pass: the gate can publish stale same-head evidence, and the required acceptance-test coverage is not implemented.
1 inline finding published.
A PR or linked Issue could change during model review without changing the PR head, allowing stale readiness evidence to be published against current state. - Re-fetch PR metadata, native closing Issues, base/head, and review threads before publication - Recompute the canonical snapshot and fail closed when it differs from reviewed evidence - Canonically sort linked Issues and test both matching and stale identity paths Generated with [Codex](https://github.com/openai)
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: 7dba682777
Reviewed range: e16db5d052..7dba682777
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: c03462ef9683b78944b5d39469e1ffa26c1836562413734afd6b605246cf2679
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 2s
Input tokens: 291,474
Cached input tokens: 242,663
Cache write tokens: 48,784
Cache hit ratio: 83.3%
Output tokens: 5,379
Reasoning tokens: 3,407
Total tokens: 296,853
Estimated Codex credits: 6.584
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:19ccc8a12eb5fb08998561ec7d34f39a1a488f1bc0405b44391cfa1c3fcd0f6c |
56s | 194,763 | 147,781 | 75.9% | 46,961 | 4,992 | 3,349 | 199,755 | 5.732 |
aggregate:c03462ef9683b78944b5d39469e1ffa26c1836562413734afd6b605246cf2679 |
6s | 96,711 | 94,882 | 98.1% | 1,823 | 387 | 58 | 97,098 | 0.852 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: FAIL
Evidence: 587cda8722b74eaf775583c0c37bb0a3f76a71fd5bcfb09271d359a54723e6db
Readiness blockers
code-review.github/workflows/openai-issue-review-dispatch.yml:92: Refresh fork PRs after linked Issue edits
One regression prevents Issue edits from invalidating readiness for fork PRs.
1 inline finding published.
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: 7dba682777
Reviewed range: e16db5d052..7dba682777
Review mode: reused
Session key: repo-1309321116-pr-11-v2
Generation key: c03462ef9683b78944b5d39469e1ffa26c1836562413734afd6b605246cf2679
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 0s
Input tokens: 0
Cached input tokens: 0
Cache write tokens: 0
Cache hit ratio: N/A
Output tokens: 0
Reasoning tokens: 0
Total tokens: 0
Estimated Codex credits: 0.000
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: FAIL
Evidence: 587cda8722b74eaf775583c0c37bb0a3f76a71fd5bcfb09271d359a54723e6db
Readiness blockers
code-review.github/workflows/openai-issue-review-dispatch.yml:92: Refresh fork PRs after linked Issue edits
One regression prevents Issue edits from invalidating readiness for fork PRs.
1 inline finding published.
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: 6069dc9a8d
Reviewed range: e16db5d052..6069dc9a8d
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: a4823bb5a470dcba9fbc62f1cb355ac9cc86deba61643318a3e8b04b5b17f5a6
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 58s
Input tokens: 464,717
Cached input tokens: 406,546
Cache write tokens: 58,135
Cache hit ratio: 87.5%
Output tokens: 4,841
Reasoning tokens: 3,074
Total tokens: 469,558
Estimated Codex credits: 7.992
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:262bf5dcae749ecb8f85f9b442534c4d63d348c5906d15e175d69a7d6beceb66 |
53s | 349,214 | 292,851 | 83.9% | 56,333 | 4,552 | 3,045 | 353,766 | 7.060 |
aggregate:a4823bb5a470dcba9fbc62f1cb355ac9cc86deba61643318a3e8b04b5b17f5a6 |
5s | 115,503 | 113,695 | 98.4% | 1,802 | 289 | 29 | 115,792 | 0.932 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: FAIL
Code review: PASS
Evidence: 15554923623118cc36fef130e989df51c0403221bb9f0ee1fa7892c3a100d59c
Readiness blockers
plan-conformanceThe linked Issue requires tests for Task-only/cross-repository/text-only linkage, stale or failed Issue evidence, plan deviations, unresolved threads, API failures, full PASS, and fail-closed behavior. The added tests cover only a small subset of metadata and snapshot cases, so the claimed readiness gate lacks the required validation coverage.
Implementation readiness is blocked by incomplete acceptance-test coverage.
No actionable findings.
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: db78097417
Reviewed range: e16db5d052..db78097417
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: f1467ed33fcd0f0baf8c486eb3f1231a617c658efadfe25be54736e58db848c8
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 6s
Input tokens: 498,157
Cached input tokens: 422,094
Cache write tokens: 76,033
Cache hit ratio: 84.7%
Output tokens: 6,098
Reasoning tokens: 2,907
Total tokens: 504,255
Estimated Codex credits: 9.679
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:8efd4ebcedce5828177a1b54cad0def1d4789d9cdd3eb32893cf3ebe5698c818 |
1m 1s | 346,906 | 273,838 | 78.9% | 73,044 | 5,745 | 2,854 | 352,651 | 8.433 |
aggregate:f1467ed33fcd0f0baf8c486eb3f1231a617c658efadfe25be54736e58db848c8 |
5s | 151,251 | 148,256 | 98.0% | 2,989 | 353 | 53 | 151,604 | 1.246 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: FAIL
Code review: PASS
Evidence: 44909af71a87988b85ad74015cfb5d38e6634350cb1776fad941da8583e7f29a
Readiness blockers
plan-conformanceThe linked Issue requires live integration PR demonstrations for invalid metadata/linkage, incomplete Issues, plan deviations, code-review failures, metadata invalidation, synchronize behavior, and a full PASS. The PR body reports only local test and lint commands, with no evidence that these required live checks were performed or otherwise satisfied.
PR readiness is blocked by missing acceptance validation evidence.
No actionable findings.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: db78097417
Reviewed range: db78097417..db78097417
Review mode: incremental
Session key: repo-1309321116-pr-11-v2
Generation key: 70114c7d525b1d726312a8440a364de09eb16c7ccf03fc68b318093c783f7547
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 10s
Input tokens: 312,413
Cached input tokens: 233,423
Cache write tokens: 78,978
Cache hit ratio: 74.7%
Output tokens: 767
Reasoning tokens: 94
Total tokens: 313,180
Estimated Codex credits: 6.683
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:0621443d3cdd98bc775c9dd9777082a70caccdc14d5b1f0eeaa4fc137b1da937 |
6s | 154,903 | 76,636 | 49.5% | 78,261 | 548 | 82 | 155,451 | 5.576 |
aggregate:70114c7d525b1d726312a8440a364de09eb16c7ccf03fc68b318093c783f7547 |
4s | 157,510 | 156,787 | 99.5% | 717 | 219 | 12 | 157,729 | 1.107 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: PASS
Evidence: f39724c370c90e7988a27feee7b789cc6f5cc223810eba34ea5820c10a4c950d
No configured readiness blockers.
No changes were introduced in this generation, and no new readiness blockers are present.
No actionable findings.
The PR reviewer should expose one authoritative required Check instead of separating execution success from Issue-led readiness in the PR Checks list. - Make OpenAI PR review fail on PR, Issue, plan, thread, or code blockers - Remove the standalone Issue dispatch and unused Issue-only review runtime - Keep model execution read-only and document the single-Check rollout Generated with [Codex](https://github.com/openai)
There was a problem hiding this comment.
❌ PR readiness: FAIL (3 blockers)
🤖 OpenAI PR review
❌ Conclusion: FAIL (2 actionable findings found)
Reviewed commit: a685485e49
Reviewed range: e16db5d052..a685485e49
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 4c4621ee02e5c96e69481d85a114e456fd50122f03e6f053caad1829d1f976d6
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 31s
Input tokens: 568,977
Cached input tokens: 497,546
Cache write tokens: 71,395
Cache hit ratio: 87.4%
Output tokens: 8,369
Reasoning tokens: 4,731
Total tokens: 577,346
Estimated Codex credits: 10.712
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:7d107ebb440215b30df906fbf626dc6bbe379fd995551b74c841a51fe9aef14a |
1m 24s | 427,298 | 359,276 | 84.1% | 67,992 | 7,773 | 4,576 | 435,071 | 9.412 |
aggregate:4c4621ee02e5c96e69481d85a114e456fd50122f03e6f053caad1829d1f976d6 |
7s | 141,679 | 138,270 | 97.6% | 3,403 | 596 | 155 | 142,275 | 1.301 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: FAIL
Code review: FAIL
Evidence: ad91a7641019349a5b41924761884662864a75d071d1604f0b278b75ee16f86b
Readiness blockers
plan-conformanceThe linked Issue requires issue-change triggers and dependent-PR recalculation. This change provides only PR/comment/manual triggers, so an Issue edit can leave a passing readiness Check unrecalculated.code-review.github/scripts/issue-review/common.mjs:76: Allow Task containers when a concrete Issue is also linkedcode-review.github/workflows/codex-openai-review.yml:351: Fail closed when a linked Issue has more than 100 sub-issues
Readiness fails due to a missing linked-Issue recalculation lifecycle and two correctness gaps.
2 inline findings published.
The unified review Check must be recalculated when its linked Issue changes, and its deterministic Issue evidence must not reject valid tracking layouts or silently truncate native relationships. - Route Issue events through the existing PR review caller - Accept Task containers when a concrete implementation Issue is also linked - Fail closed when native sub-issue evidence exceeds the query bound Generated with [Codex](https://github.com/openai)
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: 36d64978a8
Reviewed range: e16db5d052..36d64978a8
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 7711554bed64f73f0bf37531ae43b880775c165786e4003052c03fac4fc406e9
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 6s
Input tokens: 352,723
Cached input tokens: 301,225
Cache write tokens: 51,468
Cache hit ratio: 85.4%
Output tokens: 5,261
Reasoning tokens: 3,627
Total tokens: 357,984
Estimated Codex credits: 7.074
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:159f461fe2c173d256ba8272fab3a1341ff44c18d26ecd0f37861536977eeb92 |
58s | 250,521 | 200,986 | 80.2% | 49,511 | 4,924 | 3,587 | 255,445 | 6.199 |
aggregate:7711554bed64f73f0bf37531ae43b880775c165786e4003052c03fac4fc406e9 |
8s | 102,202 | 100,239 | 98.1% | 1,957 | 337 | 40 | 102,539 | 0.876 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: FAIL
Code review: PASS
Evidence: 0af75498d210dd2cb8adf11e4d0eb4508f1861fd3f6b6790bcb94c525373e39f
Readiness blockers
plan-conformanceThe linked Issue’s Code Changes Tree and acceptance criteria require issue-review/prepare.mjs, run.mjs, review-output-schema.json, and tests for design quality, malformed output, retry/checkpoint behavior, and invalid plan/acceptance criteria. This chunk adds only issue-review/common.mjs plus format-only tests, while the PR body says the Issue-only runtime was removed. Update the linked Issue with this design change or implement and validate the promised Issue-review components before readiness can pass.
The implementation omits an Issue-owned review runtime and its required coverage, diverging from the linked Issue plan.
No actionable findings.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: 36d64978a8
Reviewed range: 36d64978a8..36d64978a8
Review mode: incremental
Session key: repo-1309321116-pr-11-v2
Generation key: 1c174b19b8298415044ccae2df040931d40712c149b26d6643c1238c631ba62a
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 16s
Input tokens: 222,328
Cached input tokens: 165,207
Cache write tokens: 57,109
Cache hit ratio: 74.3%
Output tokens: 531
Reasoning tokens: 92
Total tokens: 222,859
Estimated Codex credits: 4.802
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-chunk and aggregation usage
| Turn key | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|
chunk:1/1:bdc72b0045659ae842faf12a3ce2f487e5b7a6cde5fdec4b76d0e24bf329fc80 |
7s | 108,517 | 52,125 | 48.0% | 56,386 | 362 | 92 | 108,879 | 3.986 |
aggregate:1c174b19b8298415044ccae2df040931d40712c149b26d6643c1238c631ba62a |
9s | 113,811 | 113,082 | 99.4% | 723 | 169 | 0 | 113,980 | 0.816 |
Issue-led readiness
PR contract: PASS
Issue design and plan conformance: PASS
Code review: PASS
Evidence: 2f070dc2e8836008c4e4575413d507b7c4e7ed5bafa427089979b83f6dd25ac2
No configured readiness blockers.
No changes were introduced since the last completed review, and the updated linked Issue plan matches the implemented unified review design.
No actionable findings.
Create fixed PR, Issue, and Code Review checks on each reviewed head. Cache PR metadata, each linked Issue, and code evidence independently so unchanged stages consume zero model tokens and changed snapshots send only their incremental evidence. Report per-stage cache modes and token usage in the unified native review.
There was a problem hiding this comment.
❌ PR readiness: FAIL (1 blocker)
🤖 OpenAI PR review
❌ Conclusion: FAIL (1 actionable finding found)
Reviewed commit: 8214164b4f
Reviewed range: e16db5d052..8214164b4f
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: 3dd2580b68dc3e1bd2ac262544770a14d553391487be47f0ef4110cc2a02636f
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 1m 13s
Input tokens: 568,411
Cached input tokens: 503,388
Cache write tokens: 64,969
Cache hit ratio: 88.6%
Output tokens: 5,732
Reasoning tokens: 2,943
Total tokens: 574,143
Estimated Codex credits: 9.360
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-stage token and cache usage
| Stage | Mode | Item | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|---|---|
pr |
deterministic |
pr-format:deterministic |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
pr |
full |
pr:ef46c14053314ae37cbada3631756b04e421aedd5094f5d640c4e8aad814dce5 |
5s | 19,637 | 9,250 | 47.1% | 10,381 | 268 | 90 | 19,905 | 0.807 |
issue |
full |
Issue #10 |
5s | 24,105 | 21,112 | 87.6% | 2,987 | 297 | 77 | 24,402 | 0.430 |
code |
full |
code:chunk:1/1:a0f663936cd423b799a2455f4bc778f2e597da94e26798152fb91bd1207ef80c |
59s | 422,244 | 372,508 | 88.2% | 49,700 | 4,832 | 2,747 | 427,076 | 7.249 |
code |
full |
code:aggregate:3dd2580b68dc3e1bd2ac262544770a14d553391487be47f0ef4110cc2a02636f |
4s | 102,425 | 100,518 | 98.1% | 1,901 | 335 | 29 | 102,760 | 0.873 |
Issue-led readiness
OpenAI PR Review: PASS
OpenAI Issue Review: PASS
OpenAI Code Review: FAIL
Evidence: f2f2ceb0034605dfd848055e21a1725f6e2ce9d42bf1760a120100a4f4b60e17
Readiness blockers
code-review.github/workflows/codex-openai-review.yml:435: Fail closed when a linked Issue body is truncated
The full code generation has one actionable finding: linked Issue bodies are silently truncated before readiness validation and review.
1 inline finding published.
Bind PR and Issue body truncation to canonical readiness snapshots and surface deterministic blockers before semantic or code review can pass. Use the same flags during publication-time identity verification.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: abb2225344
Reviewed range: e16db5d052..abb2225344
Review mode: full
Session key: repo-1309321116-pr-11-v2
Generation key: b2a6dd784aa73263126892f887b2d02cd5c46646bc3b8faaea3421e8b9bb57c5
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 55s
Input tokens: 367,762
Cached input tokens: 302,941
Cache write tokens: 64,779
Cache hit ratio: 82.4%
Output tokens: 4,035
Reasoning tokens: 2,338
Total tokens: 371,797
Estimated Codex credits: 7.458
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-stage token and cache usage
| Stage | Mode | Item | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|---|---|
pr |
deterministic |
pr-format:deterministic |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
pr |
full |
pr:ef46c14053314ae37cbada3631756b04e421aedd5094f5d640c4e8aad814dce5 |
8s | 29,691 | 18,872 | 63.6% | 10,810 | 548 | 148 | 30,239 | 1.000 |
issue |
full |
Issue #10 |
4s | 24,875 | 21,992 | 88.4% | 2,877 | 305 | 71 | 25,180 | 0.432 |
code |
full |
code:chunk:1/1:778dd328b482131e146aab51ab565275ddbe30e358c186528a60c5c40bd38433 |
37s | 211,637 | 163,375 | 77.2% | 48,241 | 2,980 | 2,083 | 214,617 | 5.155 |
code |
full |
code:aggregate:b2a6dd784aa73263126892f887b2d02cd5c46646bc3b8faaea3421e8b9bb57c5 |
6s | 101,559 | 98,702 | 97.2% | 2,851 | 202 | 36 | 101,761 | 0.871 |
Issue-led readiness
OpenAI PR Review: PASS
OpenAI Issue Review: PASS
OpenAI Code Review: PASS
Evidence: 4278ae7bb074742eeea6ea77d74c8f83b30ffc043186f26f2fe35c836dd80cb0
No configured readiness blockers.
The single completed chunk has no actionable findings or plan-conformance blockers. No previous findings require preservation.
No actionable findings.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: abb2225344
Reviewed range: e16db5d052..abb2225344
Review mode: reused
Session key: repo-1309321116-pr-11-v2
Generation key: b2a6dd784aa73263126892f887b2d02cd5c46646bc3b8faaea3421e8b9bb57c5
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 0s
Input tokens: 0
Cached input tokens: 0
Cache write tokens: 0
Cache hit ratio: N/A
Output tokens: 0
Reasoning tokens: 0
Total tokens: 0
Estimated Codex credits: 0.000
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-stage token and cache usage
| Stage | Mode | Item | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|---|---|
pr |
deterministic |
pr-format:deterministic |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
pr |
reused |
pr:reused |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
issue |
reused |
Issue #10 |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
code |
reused |
code:reused |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
Issue-led readiness
OpenAI PR Review: PASS
OpenAI Issue Review: PASS
OpenAI Code Review: PASS
Evidence: 4278ae7bb074742eeea6ea77d74c8f83b30ffc043186f26f2fe35c836dd80cb0
No configured readiness blockers.
The single completed chunk has no actionable findings or plan-conformance blockers. No previous findings require preservation.
No actionable findings.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: abb2225344
Reviewed range: abb2225344..abb2225344
Review mode: incremental
Session key: repo-1309321116-pr-11-v2
Generation key: b0797fec652188d0a45562033e5f4afa20df4ecc3d757f89bd8a36141256bafa
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 8s
Input tokens: 208,957
Cached input tokens: 103,826
Cache write tokens: 105,119
Cache hit ratio: 49.7%
Output tokens: 369
Reasoning tokens: 36
Total tokens: 209,326
Estimated Codex credits: 7.358
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-stage token and cache usage
| Stage | Mode | Item | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|---|---|
pr |
deterministic |
pr-format:deterministic |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
pr |
reused |
pr:reused |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
issue |
incremental |
Issue #10 |
4s | 103,443 | 51,304 | 49.6% | 52,133 | 192 | 23 | 103,635 | 3.651 |
code |
reused |
code:chunk:1:no-diff |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
code |
incremental |
code:aggregate:b0797fec652188d0a45562033e5f4afa20df4ecc3d757f89bd8a36141256bafa |
4s | 105,514 | 52,522 | 49.8% | 52,986 | 177 | 13 | 105,691 | 3.707 |
Issue-led readiness
OpenAI PR Review: PASS
OpenAI Issue Review: PASS
OpenAI Code Review: PASS
Evidence: acccad742efb325b898ffe55868e8fc05b92ee6989c479af11c30aa1cdb500ff
No configured readiness blockers.
No code changed in this generation. The prior complete review has no findings or plan-conformance blockers to preserve.
No actionable findings.
There was a problem hiding this comment.
✅ PR readiness: PASS
🤖 OpenAI PR review
✅ Conclusion: PASS
Reviewed commit: abb2225344
Reviewed range: abb2225344..abb2225344
Review mode: incremental
Session key: repo-1309321116-pr-11-v2
Generation key: 82841bd8fc8624a26dd44415b011101d20a9d423dbf08b1264b6c8f291aad509
Diff chunks: 1
Model: gpt-5.6-terra
Reasoning effort: medium
Review time: 7s
Input tokens: 215,504
Cached input tokens: 211,547
Cache write tokens: 3,945
Cache hit ratio: 98.2%
Output tokens: 328
Reasoning tokens: 7
Total tokens: 215,832
Estimated Codex credits: 1.692
Credit rate per 1M tokens: 62.5 input / 6.25 cached / 375 output
Per-stage token and cache usage
| Stage | Mode | Item | Time | Input | Cached | Cache hit | Cache write | Output | Reasoning | Total | Est. credits |
|---|---|---|---|---|---|---|---|---|---|---|---|
pr |
deterministic |
pr-format:deterministic |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
pr |
reused |
pr:reused |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
issue |
incremental |
Issue #10 |
4s | 106,892 | 104,485 | 97.7% | 2,401 | 170 | 7 | 107,062 | 0.867 |
code |
reused |
code:chunk:1:no-diff |
0s | 0 | 0 | 0.0% | 0 | 0 | 0 | 0 | 0.000 |
code |
incremental |
code:aggregate:82841bd8fc8624a26dd44415b011101d20a9d423dbf08b1264b6c8f291aad509 |
3s | 108,612 | 107,062 | 98.6% | 1,544 | 158 | 0 | 108,770 | 0.825 |
Issue-led readiness
OpenAI PR Review: PASS
OpenAI Issue Review: PASS
OpenAI Code Review: PASS
Evidence: 06d4f15a40ba5baad3b8c64195103f9cde51b06f7ff670e4ac9276e811598e84
No configured readiness blockers.
No code changed in this generation. No prior findings or plan-conformance blockers require preservation.
No actionable findings.
1 participant
Summary
OpenAI PR Review,OpenAI Issue Review, andOpenAI Code Review## 🤖 OpenAI PR reviewreport; do not create a standalone Issue workflow or per-Issue Check names@codexrequests, and manual recoveryCheck contract
OpenAI PR Reviewfails for PR format, native linkage, actionable-thread, or PR-stage blockers.OpenAI Issue Reviewaggregates deterministic and semantic design blockers from every linked Issue.OpenAI Code Reviewfails for actionable code findings or Issue-plan conformance blockers.Each Check is finalized independently as
success,failure, orcancelled. Repositories that require the full Issue-first contract should configure all three fixed names as required checks.The workflow still publishes only one native PR review. This automated result is not a GitHub approval and does not merge or deploy.
Incremental evidence
last_completed_head..current_head.Validation
node .github/scripts/issue-review/test.mjsnode .github/scripts/pr-readiness/test.mjsnode .github/scripts/pr-review/test.mjsactionlint -ignore 'property "workflow_(repository|sha)" is not defined'git diff --checkUnit coverage includes all three Check names, stage-specific verdicts, full/incremental/reused modes, per-Issue invalidation, zero-token unchanged reuse, Issue-plan revalidation, checkpoint recovery, and token/credit aggregation.
Because the new caller definition is not yet on
main, pre-merge integration usesworkflow_dispatchfrom this branch. Automaticpull_request_targetandissuesevent delivery remains a post-merge test.Closes #10