Watch CI Patterns

What it does

When your agent uses it

• Tasks that involve Containers

Example prompts

• “/watch-ci-patterns”

Requirements

• Docker
• Pre-approved tools (allowed-tools): Bash(gh run list *), Bash(gh run view *), Bash(gh workflow list *), Bash(ls .github/workflows/*), Bash(rg *), Bash(grep *), Bash(awk *), Bash(sort *), Bash(uniq *), Read, Grep

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

1. 1flake: 同じ branch の同じ commit で連続失敗 → 再 run で pass。 timing / network 由来。
2. 2drift: lockfile / version pin / contract 不整合。 pre-commit で先回り可能。
3. 3env: runner OS specific / QEMU emulation / disk space / Docker daemon。
4. 4test bug: 同じ branch で再 run しても再現。 実装変更を要する。

What it can do on your machine

Read from SKILL.md and the folder at commit 421d066. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves these tools, so the agent can use them without asking each time:

  • Bash(gh run list *)
  • Bash(gh run view *)
  • Bash(gh workflow list *)
  • Bash(ls .github/workflows/*)
  • Bash(rg *)
  • Bash(grep *)
  • Bash(awk *)
  • Bash(sort *)
  • Bash(uniq *)
  • Read

  …and 1 more on the same allowed-tools line.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • gh
  • jq
  • cargo
  • git

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md. Its commands use gh and git, which can reach the network depending on how they are called.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Watch CI Patterns loads about 1.3k tokens when it runs. Until then it costs about 73 tokens; SKILL.md has 363 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~73

When it runs · the whole SKILL.md, loaded when a task matches

~1.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from ayutaz/piper-plus at commit 421d066, republished under its MIT licence (© ayutaz). 363 words, ~1,301 tokens.

name

watch-ci-patterns

description

CI workflow の最近 run を集計し、 failure を「flake / drift / env / test bug」に自動分類する skill。 既存 `/watch-pr <PR>` の workflow-level 拡張。 引数なしで全 workflow の health check、 `--workflow <name>` で個別 workflow の最近失敗パターンを抽出。 docker-build 45% / go-ci 14% / csharp-ci 13% のような chronic flake を可視化する。

allowed-tools

Bash(gh run list *), Bash(gh run view *), Bash(gh workflow list *), Bash(ls .github/workflows/*), Bash(rg *), Bash(grep *), Bash(awk *), Bash(sort *), Bash(uniq *), Read, Grep

argument-hint

[--workflow <name>] [--limit 30] [--since 7d]

disable-model-invocation

false

gh workflow list --limit 50 --json name,state,id | jq -r '.[] | select(.state == "active") | .name'

gh run list --workflow="$WORKFLOW" --limit "$LIMIT" \
    --json conclusion,name,headBranch,createdAt,displayTitle,databaseId,event

gh run list --workflow="$WORKFLOW" --status=failure --limit 50 \
    --json headSha,headBranch,databaseId,createdAt | \
    jq -r 'group_by(.headSha) | map({sha: .[0].headSha, count: length, branches: [.[].headBranch] | unique}) | .[]'

gh run view "$RUN_ID" --log-failed 2>&1 | tail -200 > /tmp/run_$RUN_ID.log

# Pattern matching
grep -E "(disk full|no space left)" /tmp/run_$RUN_ID.log && echo "ENV: disk"
grep -E "(connection refused|timeout|TLS)" /tmp/run_$RUN_ID.log && echo "ENV: network"
grep -E "(go.mod|inconsistent vendoring|missing go.sum)" /tmp/run_$RUN_ID.log && echo "DRIFT: go.mod"
grep -E "(ruff|format.*drift|cargo fmt)" /tmp/run_$RUN_ID.log && echo "DRIFT: format"
grep -E "(qemu|exec format error|illegal instruction)" /tmp/run_$RUN_ID.log && echo "ENV: arm64-emulation"
grep -E "FAILED.*test_" /tmp/run_$RUN_ID.log | head -5  # 実 test failure 名

## CI Pattern Report — $(date +%Y-%m-%d)

### Summary
- workflows scanned: $N
- total runs (last $LIMIT): $TOTAL
- failure rate (overall): $RATE%

### Top failing workflows
| Workflow | Failure% | Primary category |
|----------|---------|------------------|
| docker-build | 45% | env (arm64 QEMU) |
| go-ci | 14% | drift (go.mod) |
| ... | ... | ... |

### Action items
- [ ] go-ci drift: `go mod tidy` を pre-push に追加検討
- [ ] docker-build flake: arm64 build を concurrency 制限
- [ ] csharp-ci issue-426: 個別 fix (test bug)

# 全 workflow の health check
/watch-ci-patterns

# docker-build に絞って深堀り
/watch-ci-patterns --workflow docker-build.yml --limit 50

# 最近 24h の failure のみ
/watch-ci-patterns --since 24h

# loop で 30 分ごと監視
/loop 30m /watch-ci-patterns --since 24h

© ayutaz, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/watch-ci-patterns of ayutaz/piper-plus.

Open the folder on GitHubat commit 421d066