From 817891cf8a9a636e9fa324628cb05938129fd49f Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 31 Aug 2026 11:22:32 +0000 Subject: [PATCH 1/3] fix(ci): remove stale hardcoded timeout-minutes claim from sidecar comment The gateway-preflight comment in contextual_orchestrator_review_sidecar.sh claimed "this required-workflow job already budgets 120 minutes (see timeout-minutes in strix.yml/noema-review.yml)". Both halves of that claim are now stale: strix.yml's job timeout-minutes is 200, not 120, and noema-review.yml had no timeout-minutes at all until PR #1509 gave it timeout-minutes: 300 (fixing a bug where a slow LLM review could outlive the GitHub App token used to submit its verdict). Reworded to reference each caller workflow's own job-level timeout-minutes without pinning a shared number, since the two workflows' budgets now differ and can change independently without re-staling this comment again. Comment-only change; no behavior, test, or hash-pinned file touched. Co-Authored-By: Claude Sonnet 5 Claude-Session: https://claude.ai/code/session_015Gs7KmNvH75nxz1sL8mKjw --- scripts/ci/contextual_orchestrator_review_sidecar.sh | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/scripts/ci/contextual_orchestrator_review_sidecar.sh b/scripts/ci/contextual_orchestrator_review_sidecar.sh index e4984f643b..754ef5078f 100755 --- a/scripts/ci/contextual_orchestrator_review_sidecar.sh +++ b/scripts/ci/contextual_orchestrator_review_sidecar.sh @@ -428,8 +428,10 @@ printf '{"model":"%s","messages":[{"role":"system","content":"You are a helpful # in 18s, then this identical request against that same healthy route being # cut off by curl's own timeout at exactly 30.0s -- "gateway preflight # request could not reach the local sidecar" is this curl failure, not an -# actual connectivity problem. This required-workflow job already budgets -# 120 minutes (see timeout-minutes in strix.yml/noema-review.yml), and the +# actual connectivity problem. Each caller workflow already budgets its own +# multi-hour job-level timeout-minutes for this required-workflow job (see +# strix.yml and noema-review.yml -- the two currently differ and either can +# change independently, so no specific number is pinned here), and the # org's own stated policy accepts multi-hour central review latency in # favor of accuracy over speed -- a 30s bound on one preflight self-check # contradicted that policy and rejected a route the routing probe had just From a463b7ec2ffb9d1cb236cf0a68ed3e5d1453b1b2 Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 1 Sep 2026 00:31:08 +0000 Subject: [PATCH 2/3] fix(ci): correct sidecar comment's noema-review.yml timeout claim Devin Review (analysis finding on #1510) caught that the replacement comment still claimed noema-review.yml budgets its own job-level timeout-minutes. It does not: PR #1509, which the prior PR body credited with adding timeout-minutes: 300 to that job, never actually made that change (its own body states so explicitly) and was closed unmerged. noema-review.yml's `noema-review` job has no job-level timeout-minutes at all, so its real outer bound is GitHub Actions' own 360-minute default -- not a number this repo sets. Reworded the comment to state that accurately instead of claiming a value that doesn't exist. --- Generated by Claude Code --- scripts/ci/contextual_orchestrator_review_sidecar.sh | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/scripts/ci/contextual_orchestrator_review_sidecar.sh b/scripts/ci/contextual_orchestrator_review_sidecar.sh index 754ef5078f..f5f3fdb07f 100755 --- a/scripts/ci/contextual_orchestrator_review_sidecar.sh +++ b/scripts/ci/contextual_orchestrator_review_sidecar.sh @@ -428,10 +428,12 @@ printf '{"model":"%s","messages":[{"role":"system","content":"You are a helpful # in 18s, then this identical request against that same healthy route being # cut off by curl's own timeout at exactly 30.0s -- "gateway preflight # request could not reach the local sidecar" is this curl failure, not an -# actual connectivity problem. Each caller workflow already budgets its own -# multi-hour job-level timeout-minutes for this required-workflow job (see -# strix.yml and noema-review.yml -- the two currently differ and either can -# change independently, so no specific number is pinned here), and the +# actual connectivity problem. The calling job's own outer bound already +# gives multi-hour headroom for this required-workflow job: strix.yml's +# job-level timeout-minutes is 200 (can change independently of this +# value, so no specific number is pinned here); noema-review.yml's +# `noema-review` job sets no job-level timeout-minutes at all, so its +# outer bound is GitHub Actions' own 360-minute default. Either way, the # org's own stated policy accepts multi-hour central review latency in # favor of accuracy over speed -- a 30s bound on one preflight self-check # contradicted that policy and rejected a route the routing probe had just From 58af78284dcfe939b7dead6088bd15befc667d9f Mon Sep 17 00:00:00 2001 From: Claude Date: Tue, 1 Sep 2026 00:34:03 +0000 Subject: [PATCH 3/3] fix(ci): cover pr-review-autofix.yml as a third timeout-less sidecar caller Codex Review caught that the prior fix (61d1de3f) only accounted for strix.yml and noema-review.yml, but pr-review-autofix.yml also invokes this sidecar script (scripts/ci/contextual_orchestrator_review_sidecar.sh line 259) and also sets no job-level timeout-minutes -- confirmed via grep against the actual workflow file. Reworded the comment to name all three known callers accurately and, since this comment has now gone stale twice on an incomplete caller inventory, added an explicit note not to trust this list's completeness/durability without checking the caller workflows directly. --- Generated by Claude Code --- .../ci/contextual_orchestrator_review_sidecar.sh | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/scripts/ci/contextual_orchestrator_review_sidecar.sh b/scripts/ci/contextual_orchestrator_review_sidecar.sh index f5f3fdb07f..fc5686cf26 100755 --- a/scripts/ci/contextual_orchestrator_review_sidecar.sh +++ b/scripts/ci/contextual_orchestrator_review_sidecar.sh @@ -429,12 +429,15 @@ printf '{"model":"%s","messages":[{"role":"system","content":"You are a helpful # cut off by curl's own timeout at exactly 30.0s -- "gateway preflight # request could not reach the local sidecar" is this curl failure, not an # actual connectivity problem. The calling job's own outer bound already -# gives multi-hour headroom for this required-workflow job: strix.yml's -# job-level timeout-minutes is 200 (can change independently of this -# value, so no specific number is pinned here); noema-review.yml's -# `noema-review` job sets no job-level timeout-minutes at all, so its -# outer bound is GitHub Actions' own 360-minute default. Either way, the -# org's own stated policy accepts multi-hour central review latency in +# gives multi-hour headroom for this required-workflow job regardless of +# which caller invokes this script: strix.yml's job-level timeout-minutes +# is 200; noema-review.yml's `noema-review` job and pr-review-autofix.yml's +# job that runs this script both set no job-level timeout-minutes at all, +# so their outer bound is GitHub Actions' own 360-minute default. Do not +# assume this list of callers or their timeout values is exhaustive or +# durable -- check each caller workflow directly rather than trusting this +# comment if it matters at the time of reading. Either way, the org's own +# stated policy accepts multi-hour central review latency in # favor of accuracy over speed -- a 30s bound on one preflight self-check # contradicted that policy and rejected a route the routing probe had just # proven healthy. 120s keeps this a bounded, fail-closed check while giving