Diagnostics Rules & levers Pipelines Executions & journal Trace table Metrics ModelsMonths: 2026-11 2026-10

Diagnostics

Problems detected by fixed rules over 479 recorded run(s) and 153 work item(s) — no LLM involved. Each finding separates what was observed from a hypothesis about it, and lists the runs it was computed from.

Machine-readable copy: findings.json — the 8 finding(s) below, over 479 recorded run(s), generated 2026-11-01T22:13:24.172Z. It is the input of /ai-dev-tools:loop-insights (Rules and levers).

Category0 critical · 3 warning · 5 info

warningReview

PRs not approved after several reviews

Affects 61 run(s) · 13 item(s)Since 2026-10-16 23:01 UTCactive · last seen 2026-10-19 17:45 UTC

Observed: 13 PR(s) have their last 3+ verdicts without an APPROVE: #218 (3), #219 (3), #220 (3), #221 (3), #222 (3), #225 (3), #226 (5), #227 (3), #231 (3), #234 (3), #235 (4), #241 (11), #242 (14).
Impact: 61 review run(s) without converging
Frequency: 13 PR(s) · 61 run(s) affected

Evidence · 25 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-16 23:02	38979116693-1	#218	`review.verdict`	verdict=WITHHELD
2026-10-16 23:42	38982767337-1	#218	`review.verdict`	verdict=WITHHELD
2026-10-17 00:19	38985736790-1	#219	`review.verdict`	verdict=WITHHELD
2026-10-17 00:20	38985775751-1	#219	`review.verdict`	verdict=REQUEST_CHANGES
2026-10-17 00:43	38987589546-1	#219	`review.verdict`	verdict=REQUEST_CHANGES
2026-10-17 01:20	38990364970-1	#220	`review.verdict`	verdict=WITHHELD
2026-10-17 01:22	38990406522-1	#220	`review.verdict`	verdict=WITHHELD
2026-10-17 01:51	38992537605-1	#220	`review.verdict`	verdict=WITHHELD
2026-10-17 02:34	38995413070-1	#221	`review.verdict`	verdict=REQUEST_CHANGES
2026-10-17 02:44	38996052770-1	#221	`review.verdict`	verdict=WITHHELD
2026-10-17 02:48	38996292639-1	#221	`review.verdict`	verdict=WITHHELD
2026-10-17 02:49	38996318417-1	#222	`review.verdict`	verdict=WITHHELD
2026-10-17 02:50	38996353300-1	#222	`review.verdict`	verdict=WITHHELD
2026-10-17 14:07	39039318413-1	#222	`review.verdict`	verdict=WITHHELD
2026-10-19 03:29	39194824561-1	#225	`review.verdict`	verdict=REQUEST_CHANGES
2026-10-19 03:44	39196061218-1	#225	`review.verdict`	verdict=WITHHELD
2026-10-19 04:09	39197996406-1	#225	`review.verdict`	verdict=WITHHELD
2026-10-19 04:46	39201050992-1	#226	`review.verdict`	verdict=WITHHELD
2026-10-19 11:52	39247627963-1	#226	`review.verdict`	verdict=WITHHELD
2026-10-19 11:58	39248382282-1	#226	`review.verdict`	verdict=WITHHELD
2026-10-19 14:06	39265779193-1	#227	`review.verdict`	verdict=WITHHELD
2026-10-19 14:07	39265836758-1	#227	`review.verdict`	verdict=WITHHELD
2026-10-19 14:52	39272705384-1	#227	`review.verdict`	verdict=WITHHELD
2026-10-19 17:45	39296748777-1	#231	`review.verdict`	verdict=WITHHELD
2026-10-16 23:01	38979056167-1	#218	`review.verdict`	verdict=WITHHELD

First 25 entries; 61 run(s) affected.

Investigate →

warningCode generation

Code generation is mostly blocked

Affects 17 run(s)Since 2026-10-19 03:08 UTCactive · last seen 2026-10-31 01:26 UTC

Observed

17 of 18 generation run(s) produced no patch (source: model 12, unspecified 2, denylist 3); 1 completed. Blocked by design (protected or withheld targets): 5; anomaly: 12 (the model blocked the issue, or the rejection is not a design rule).

Impact

17 issue(s) escalated instead of getting a PR

Frequency

17 run(s) · 17 run(s) affected

Hypothesis

A model-sourced block usually means the files to change were withheld from its context (size or protected paths); the escalation comment on the issue says which.

Suggested lever

Code generation is blocked before it proposes a patch

config/models.yaml#generation_max_input_tokens · no eval suite measures it: verify by hand and say so in the PR
prompts/generation-system.md · no eval suite measures it: verify by hand and say so in the PR
prompts/validation-system.md · run the live evals.yml (suite validation)

Stage generation · ADR-0029 · every change also requires docs-and-tests; the lever says where to look, not what caused the finding (this rule's entry).

Draft a proposal (read-only, nothing is posted or created): /ai-dev-tools:loop-insights --finding codegen-blocked

Evidence · 17 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-31 01:26	41155321170-1	—	`code_gen.skipped`	reason=blocked
2026-10-31 01:19	41154713645-1	—	`code_gen.skipped`	reason=guardrail_rejected · rules=protected_path · error=AI response changes[0] target_path "scripts/audit-branches.js" is in a protected path (scripts/)
2026-10-31 01:18	41154652610-1	—	`code_gen.skipped`	reason=blocked
2026-10-30 19:09	41112867211-1	—	`code_gen.skipped`	reason=blocked
2026-10-30 18:39	41108652433-1	—	`code_gen.skipped`	reason=blocked
2026-10-30 14:41	41073850373-1	—	`code_gen.skipped`	reason=blocked
2026-10-28 19:34	40755480973-1	—	`code_gen.skipped`	reason=blocked
2026-10-28 19:31	40754964443-1	—	`code_gen.skipped`	reason=blocked
2026-10-28 19:30	40754918718-1	—	`code_gen.skipped`	reason=blocked
2026-10-28 19:21	40753415508-1	—	`code_gen.skipped`	reason=blocked
2026-10-21 04:08	39542338480-1	—	`code_gen.skipped`	reason=guardrail_rejected · rules=protected_path · error=AI response changes[0] target_path "docs/evals-review-anthropic-comparison.md" is in a protected path (docs/)
2026-10-20 01:38	39354513597-1	—	`code_gen.skipped`	reason=blocked
2026-10-19 03:10	39193378100-1	—	`code_gen.skipped`	reason=blocked
2026-10-19 03:09	39193267771-1	—	`code_gen.skipped`	reason=blocked
2026-10-19 03:08	39193237010-1	—	`code_gen.skipped`	reason=blocked
2026-10-19 03:08	39193225895-1	—	`code_gen.skipped`	reason=blocked
2026-10-19 03:08	39193212668-1	—	`code_gen.skipped`	reason=blocked

Investigate →

warningLLM provider

Frequent provider fallbacks

Affects 18 run(s)Since 2026-10-16 04:07 UTCactive · last seen 2026-10-31 01:26 UTC

Observed

18 fallback(s) for 354 LLM call(s): groq → anthropic (UNKNOWN) ×11, groq → anthropic (rate_limit_exceeded) ×5, groq → anthropic (json_validate_failed) ×1, groq → anthropic (TRANSIENT) ×1. 11 of them carry no classified error type: their cause is not recorded.

Impact

Answers come from the fallback model, at its price

Frequency

18 fallback(s) · 18 run(s) affected

Suggested lever

The primary provider fails often enough that the fallback answers

AI_PROVIDER_<STAGE> · no eval suite measures it: verify by hand and say so in the PR
retry policy (ADR-0022) · no eval suite measures it: verify by hand and say so in the PR
provider tier · no eval suite measures it: verify by hand and say so in the PR

Stage provider · ADR-0022, ADR-0025 · every change also requires docs-and-tests; the lever says where to look, not what caused the finding (this rule's entry).

Draft a proposal (read-only, nothing is posted or created): /ai-dev-tools:loop-insights --finding llm-fallbacks

Evidence · 18 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-31 01:26	41155321170-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-31 01:18	41154652610-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-30 19:09	41112867211-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-30 18:39	41108652433-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-30 14:41	41073850373-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-28 19:31	40754964443-1	—	`llm_fallback`	error_type=json_validate_failed · status=400
2026-10-28 19:20	40753415508-1	—	`llm_fallback`	error_type=UNKNOWN
2026-10-20 17:21	39465136002-1	#242	`llm_fallback`	error_type=rate_limit_exceeded · status=429
2026-10-20 17:14	39464169942-1	#242	`llm_fallback`	error_type=rate_limit_exceeded · status=429
2026-10-20 17:09	39463376417-1	#242	`llm_fallback`	error_type=rate_limit_exceeded · status=429
2026-10-20 03:11	39362880138-1	#241	`llm_fallback`	error_type=rate_limit_exceeded · status=429
2026-10-20 02:47	39360679664-1	#240	`llm_fallback`	error_type=rate_limit_exceeded · status=429
2026-10-20 01:44	39355091245-1	#238	`llm_fallback`	error_type=UNKNOWN
2026-10-19 03:10	39193378100-1	—	`llm_fallback`	error=groq: generation prompt is ~3532 estimated input tokens, over generation_max_input_tokens (3500) in config/models.yaml: Groq would reject it with 413 (one request > 8K TPM). Shorten the input or raise… · error_type=UNKNOWN
2026-10-19 03:08	39193237010-1	—	`llm_fallback`	error=groq: generation prompt is ~7488 estimated input tokens, over generation_max_input_tokens (3500) in config/models.yaml: Groq would reject it with 413 (one request > 8K TPM). Shorten the input or raise… · error_type=UNKNOWN
2026-10-19 03:08	39193225895-1	—	`llm_fallback`	error=groq: generation prompt is ~5590 estimated input tokens, over generation_max_input_tokens (3500) in config/models.yaml: Groq would reject it with 413 (one request > 8K TPM). Shorten the input or raise… · error_type=UNKNOWN
2026-10-19 03:08	39193212668-1	—	`llm_fallback`	error=groq: generation prompt is ~6593 estimated input tokens, over generation_max_input_tokens (3500) in config/models.yaml: Groq would reject it with 413 (one request > 8K TPM). Shorten the input or raise… · error_type=UNKNOWN
2026-10-16 04:07	38848182099-1	#217	`llm_fallback`	error=groq: Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (T… · error_type=TRANSIENT

Investigate →

infoReview

Reviews withhold their verdict — resolved since 2026-10-23

Affects 67 run(s) · 19 item(s)Since 2026-10-16 23:01 UTCresolved · last seen 2026-10-23 14:36 UTC

Observed: 67 of 271 verdicts are WITHHELD (25 %); 66 of them had code files not reviewed, 56 would have been APPROVE (withheld_from). They cover 53 PR head(s) on 19 PR(s). Guardrail: 61 (code cut, ADR-0038: 61; CI evidence, ADR-0026: 0); anomaly: 6 (REQUEST_CHANGES on unseen code: 0; adversarial protocol not met, ADR-0049: 1; unknown cause: 0; fallback answered a prompt cut to the primary's budget: 5). Answered by: groq 61, anthropic (fallback) 5, anthropic 1; input budget 6300 tokens.
Impact: 67 review(s) ended without a usable verdict; a person has to review
Frequency: 67 verdict(s) on 53 head(s) of 19 PR(s) · 67 run(s) affected
Hypothesis: The guardrail worked as designed: approval is withheld on code the review did not see (ADR-0038). The lever is the provider or its input budget (review_max_input_tokens, ADR-0025), not the review rule.

Evidence · 25 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-23 14:36	39970922373-1	#282	`review.verdict`	verdict=WITHHELD
2026-10-20 17:21	39465136002-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 17:14	39464169942-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 17:09	39463376417-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 17:02	39462591400-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:54	39461426672-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:45	39460097943-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:39	39459145806-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:32	39458198456-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:27	39457410857-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:21	39456552143-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 16:14	39455474898-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 15:48	39451632457-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 13:41	39432222887-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 13:39	39432164291-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 11:31	39415443047-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 11:26	39414840580-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 11:18	39413913587-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 11:13	39413251169-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 11:07	39412599427-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 11:00	39411739301-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 10:53	39410929604-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 10:46	39410173183-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 10:40	39409528303-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-16 23:01	38979056167-1	#218	`review.verdict`	verdict=WITHHELD

First 25 entries; 67 run(s) affected.

Investigate →

infoReview

Same PR head reviewed more than once — resolved since 2026-10-23

Affects 32 run(s) · 16 item(s)Since 2026-10-16 23:01 UTCresolved · last seen 2026-10-23 14:38 UTC

Observed: 16 PR head(s) got 16 extra review run(s) (triggers: pull_request, push); 2 of them got different verdicts on the same code. Guardrail: 16 (after a WITHHELD verdict, ADR-0028: 16; manual re-run: 0); anomaly: 0 (two runs on one head, none skipped by the marker).
Impact: 16 redundant review run(s) and their LLM calls
Frequency: 16 head(s) · 32 run(s) affected

Evidence · 25 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-23 14:36	39970922373-1	#282	`review.verdict`	verdict=WITHHELD
2026-10-23 14:38	39970949395-1	#282	`review.verdict`	verdict=APPROVE
2026-10-20 13:39	39432164291-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 13:41	39432222887-1	#242	`review.verdict`	verdict=WITHHELD
2026-10-20 03:10	39362802426-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 03:11	39362880138-1	#241	`review.verdict`	verdict=WITHHELD
2026-10-20 02:46	39360606464-1	#240	`review.verdict`	verdict=WITHHELD
2026-10-20 02:47	39360679664-1	#240	`review.verdict`	verdict=WITHHELD
2026-10-20 01:48	39355352614-1	#239	`review.verdict`	verdict=WITHHELD
2026-10-20 01:49	39355384135-1	#239	`review.verdict`	verdict=WITHHELD
2026-10-20 01:14	39352232883-1	#235	`review.verdict`	verdict=WITHHELD
2026-10-20 01:15	39352275774-1	#235	`review.verdict`	verdict=WITHHELD
2026-10-19 23:43	39343217446-1	#234	`review.verdict`	verdict=WITHHELD
2026-10-19 23:45	39343261506-1	#234	`review.verdict`	verdict=WITHHELD
2026-10-19 23:19	39340743028-1	#233	`review.verdict`	verdict=WITHHELD
2026-10-19 23:20	39340808744-1	#233	`review.verdict`	verdict=WITHHELD
2026-10-19 17:45	39296748777-1	#231	`review.verdict`	verdict=WITHHELD
2026-10-19 17:47	39296827272-1	#231	`review.verdict`	verdict=WITHHELD
2026-10-19 15:16	39276335482-1	#228	`review.verdict`	verdict=WITHHELD
2026-10-19 15:18	39276410959-1	#228	`review.verdict`	verdict=WITHHELD
2026-10-19 14:06	39265779193-1	#227	`review.verdict`	verdict=WITHHELD
2026-10-19 14:07	39265836758-1	#227	`review.verdict`	verdict=WITHHELD
2026-10-19 04:34	39200119948-1	#226	`review.verdict`	verdict=WITHHELD
2026-10-19 04:36	39200139344-1	#226	`review.verdict`	verdict=WITHHELD
2026-10-16 23:01	38979056167-1	#218	`review.verdict`	verdict=WITHHELD

First 25 entries; 32 run(s) affected.

Investigate →

infoReview

PRs needed several external review rounds — resolved since 2026-10-26

Affects 1 run(s) · 1 item(s)Since 2026-10-25 13:36 UTCresolved · last seen 2026-10-25 13:35 UTC

Observed: 1 of 34 PR(s) with a recorded external review had 3+ rounds (#296: 9 rounds, 13 threads, 0 P0/P1); they hold 13 of the 40 external threads recorded. Counts are a lower bound: the reviewer answers a head after the run that recorded them.
Impact: 9 external review round(s) on 1 PR(s)
Frequency: 1 PR(s) · 1 run(s) affected
Hypothesis: Each round may find a neighbour of the previous fix (same defect class), which a fix that sweeps the class would remove; the records hold counts, not the findings, so this is not verified.

Evidence · 1 entry

Time (UTC)	Run	PR	Event	Recorded detail	Source
2026-10-25 13:36	40198235010-1	#296	`review.external`	—

Investigate →

infoAuto-fix

Auto-fix blocked by design (ADR-0021, ADR-0029)

Affects 21 run(s) · 13 item(s)Since 2026-10-17 00:20 UTCactive · last seen 2026-10-30 18:58 UTC

Observed: 2 delivered push(es) for 21 attempt(s); skip reasons: no_changes 14, self_modification 1, guardrail_rejected 4. Blocked by design: 18 (files withheld by budget 14, protected or unshown target 4); anomaly: 1.
Impact: 19 attempt(s) ended without a change, usually escalated to a human
Frequency: 19 of 21 attempt(s) · 21 run(s) affected
Hypothesis: The guardrails worked as designed: the attempts stopped on a protected or unshown target, or on files the budget withheld from the prompt. The lever is the input budget (autofix_max_input_tokens, ADR-0017) or a human fix, not the guardrail.

Evidence · 19 entries

Time (UTC)	Run	PR	Event	Recorded detail	Source
2026-10-30 18:59	41111383676-1	#322	`autofix.skipped`	reason=no_changes
2026-10-28 19:33	40755215086-1	#322	`autofix.skipped`	reason=no_changes
2026-10-27 18:48	40563001144-1	#312	`autofix.skipped`	reason=no_changes
2026-10-26 15:36	40350544084-1	#311	`autofix.skipped`	reason=self_modification	GitHub
2026-10-26 00:52	40251235102-1	#307	`autofix.skipped`	reason=no_changes
2026-10-25 12:42	40194269858-1	#297	`autofix.skipped`	reason=guardrail_rejected · rules=protected_path
2026-10-25 03:44	40159225402-1	#297	`autofix.skipped`	reason=guardrail_rejected · rules=protected_path
2026-10-22 19:42	39842401383-1	#267	`autofix.skipped`	reason=no_changes
2026-10-22 19:36	39841571957-1	#267	`autofix.skipped`	reason=no_changes
2026-10-21 03:46	39540372924-1	#256	`autofix.skipped`	reason=no_changes
2026-10-21 03:08	39536847321-1	#254	`autofix.skipped`	reason=no_changes
2026-10-21 03:03	39536378113-1	#254	`autofix.skipped`	reason=no_changes
2026-10-21 02:31	39533424594-1	#253	`autofix.skipped`	reason=no_changes
2026-10-20 23:26	39514964984-1	#246	`autofix.skipped`	reason=guardrail_rejected · rules=protected_path
2026-10-20 23:23	39514660585-1	#246	`autofix.skipped`	reason=no_changes
2026-10-19 03:30	39194915673-1	#225	`autofix.skipped`	reason=guardrail_rejected · rules=protected_path
2026-10-17 02:35	38995470592-1	#221	`autofix.skipped`	reason=no_changes
2026-10-17 00:43	38987671334-1	#219	`autofix.skipped`	reason=no_changes
2026-10-17 00:20	38985909027-1	#219	`autofix.skipped`	reason=no_changes

First 19 entries; 21 run(s) affected.

Investigate →

infoLLM provider

Time lost waiting on rate limits — resolved since 2026-10-19

Affects 6 run(s)Since 2026-10-16 03:59 UTCresolved · last seen 2026-10-19 03:08 UTC

Observed: 28 retry wait(s) on HTTP 429, totalling 12.3 min across 6 run(s).
Impact: 12.3 min of run time spent waiting
Frequency: 28 wait(s) · 6 run(s) affected

Evidence · 25 entries

Time (UTC)	Run	PR	Event	Recorded detail
2026-10-19 03:08	39193218816-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193218816-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:09	39193218816-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:09	39193218816-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193212735-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193212735-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193208871-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:08	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:09	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-19 03:09	39193204024-1	—	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-17 02:48	38996318417-1	#222	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-17 02:49	38996318417-1	#222	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:00	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:01	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:01	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:02	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:03	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:04	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:04	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 04:05	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429
2026-10-16 03:59	38848182099-1	#217	`retry.wait`	error=Groq API HTTP error 429: {"error":{"message":"Rate limit reached for model `openai/gpt-oss-120b` in organization `org_01kp6a0j2bfzercy319x9ygzf5` service tier `on_demand` on tokens per minute (TPM): L… · status=429

Investigate →

Data gaps · what the records cannot tell

insufficient data Merge is journaled only from item-closed.yml on
A merge or a closing is recorded (item.closed) only for items closed after item-closed.yml existed, and only in this repository; the Pipelines page asks GitHub for merged_at in the browser for the rest.
insufficient data Closings before the closing journal are not recorded
Only items closed or merged after item-closed.yml was merged have an item.closed event. An item closed before it that last escalated still reads as escalated: the dashboard does not infer a closing from GitHub or from silence. The Pipelines page re-checks pull requests live.
insufficient data Running state is not observable
A record is written when its run completes, so a run in progress never appears; the latest state of an item can be older than GitHub.
insufficient data Runs outside any work-item trace
23 run(s) have no trace index: they appear in Investigation but not in Pipelines.
insufficient data Token counts are estimates
85 of 354 LLM call(s) carry chars/4 estimates, not provider-measured usage: their cost excludes thinking and prompt cache.
insufficient data Duration trend not computable
Fewer than 3 runs in one of the two latest 7-day periods for: pr_prepare. No degradation is reported for them — that is not the same as no degradation.

Rules and thresholds

A pattern is reported only above these thresholds (RULES in scripts/lib/pipeline_diagnostics.mjs); below them it is not separated from noise. Which setting governs each rule: Rules and levers.

`minVerdicts`	3
`withheldShare`	0.25
`truncatedShare`	0.5
`reviewStreak`	3
`autofixMinAttempts`	2
`autofixNoPushShare`	0.5
`conflictsMinAttempts`	2
`conflictsUnresolvedShare`	0.5
`fallbackMin`	2
`durationMinRuns`	3
`durationRatio`	1.5
`stageFailures`	2
`externalRounds`	3
`activeWindow`	10

Diagnostics

Running with issues: 3 problem(s) to look at

Data gaps · what the records cannot tell