Move CI failure diagnosis docs into ci-fails-buildkite skill (#45975)

Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com>

Vadim Gimpelson committed Jun 20, 2026 at 00:12 UTC ca7e1f2c43834d1e720b7377e2832097978c1e35
4 files changed +40 -15
.claude/skills/ci-fails-buildkite/SKILL.md new
+35
@@ -0,0 +1,35 @@
1 +---
2 +name: ci-fails-buildkite
3 +description: Fetch and diagnose vLLM Buildkite CI failure logs. Use when investigating failing CI jobs on a PR or build, when the user pastes a buildkite.com URL, or asks to fetch/diagnose CI logs.
4 +---
5 +
6 +# Diagnosing vLLM Buildkite CI Failures
7 +
8 +Buildkite logs are public; no login needed.
9 +
10 +`.buildkite/scripts/ci-fetch-log.sh` saves each log as `ci-<build>-<job-name>.log`, stripped of timestamps and ANSI codes. Existing files are kept; set `CI_FETCH_LOG_FORCE=1` to refetch.
11 +
12 +## Fetching logs
13 +
14 +```bash
15 +# All failed jobs in a PR's latest build (current branch's PR if omitted):
16 +.buildkite/scripts/ci-fetch-log.sh --pr <PR>
17 +
18 +# All failed jobs in a build (--soft also includes soft-failed jobs;
19 +# --all fetches every finished job):
20 +.buildkite/scripts/ci-fetch-log.sh "https://buildkite.com/vllm/ci/builds/<N>"
21 +
22 +# One job — `gh pr checks` URLs (#<job_uuid>) and web UI URLs (?sid=) both
23 +# work; pass "-" as a second argument to stream to stdout:
24 +.buildkite/scripts/ci-fetch-log.sh "https://buildkite.com/vllm/ci/builds/<N>#<job_uuid>"
25 +```
26 +
27 +To clean an already-downloaded log with `.buildkite/scripts/ci-clean-log.sh`:
28 +
29 +```bash
30 +./ci-clean-log.sh ci.log
31 +```
32 +
33 +## Reference
34 +
35 +See [docs/contributing/ci/failures.md](../../../docs/contributing/ci/failures.md) for the full guide: filing CI failure issues, investigating/bisecting, reproducing flaky tests, and daily triage.
.github/CODEOWNERS
+2 -3
@@ -2,15 +2,14 @@
2 # for more info about CODEOWNERS file
3
4 # This lists cover the "core" components of vLLM that require careful review
5 -/vllm/compilation @zou3519 @youkaichao @ProExpertProg @BoyuanFeng @vadiklyutiy
5 +/vllm/compilation @zou3519 @youkaichao @ProExpertProg @BoyuanFeng
6 /vllm/distributed/kv_transfer @NickLucche @ApostaC @orozery @xuechendi
7 /vllm/lora @jeejeelee
8 /vllm/model_executor/layers/attention @LucasWilkinson @MatthewBonanni
9 /vllm/model_executor/layers/fused_moe @mgoin @pavanimajety @zyongye
10 /vllm/model_executor/layers/quantization @mgoin @robertgshaw2-redhat @tlrmchlsmth @yewentao256 @pavanimajety @zyongye
11 /vllm/model_executor/layers/mamba @tdoublep @tomeras91
12 -/vllm/model_executor/layers/mamba/gdn_linear_attn.py @tdoublep @ZJY0516 @vadiklyutiy
13 -/vllm/model_executor/layers/rotary_embedding.py @vadiklyutiy
12 +/vllm/model_executor/layers/mamba/gdn/qwen_gdn_linear_attn.py @tdoublep @ZJY0516 @vadiklyutiy
13 /vllm/model_executor/model_loader @22quinn
14 /vllm/model_executor/layers/batch_invariant.py @yewentao256
15 /vllm/ir @ProExpertProg
.gitignore
+3 -1
@@ -199,7 +199,9 @@ cython_debug/
199 .vscode/
200
201 # Claude
202 -.claude/
202 +.claude/*
203 +!.claude/skills/
204 +!.claude/skills/**
205
206 # Codex
207 .codex/
AGENTS.md
-11
@@ -114,17 +114,6 @@ Follow these rules for all code changes in this repository:
114 - Keep comments and docstrings minimal and concise.
115 - Assume the reader is familiar with vLLM.
116
117 -### Diagnosing CI failures
118 -
119 -Buildkite logs are public; no login needed. Details: [docs/contributing/ci/failures.md](docs/contributing/ci/failures.md).
120 -
121 -```bash
122 -# All failed-job logs for a PR's latest build (current branch's PR if omitted):
123 -.buildkite/scripts/ci-fetch-log.sh --pr <PR>
124 -# Any Buildkite build or job URL also works:
125 -.buildkite/scripts/ci-fetch-log.sh "<buildkite_url>"
126 -```
127 -
117 ### Commit messages
118
119 Add attribution using commit trailers such as `Co-authored-by:` (other projects use `Assisted-by:` or `Generated-by:`). For example: