メニュー
Guidance for using github actions. Use when writing github actions and workflows.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Control herdr from inside it. Manage workspaces and tabs, split panes, spawn agents, read output, and wait for state changes — all via CLI commands that talk to the running herdr instance over a local unix socket. Use when running inside herdr (HERDR_ENV=1).
Analyze pi session token usage and produce a breakdown report. Sums the usage blocks recorded in pi session logs. Use when the user wants a token report, usage breakdown, "how much have I used", cached-vs-fresh token split, per-model or per-project cost/tokens, or asks to analyze pi usage.
Use when using the agent tool or delegating to subagents. Use it for tasks that benefit from isolated context: codebase exploration, planning, focused implementation, or code review.
Ask which skill or flow fits your situation. A router over the user-invoked skills in this repo.
Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.
Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.
| name | github-actions |
| description | Guidance for using github actions. Use when writing github actions and workflows. |
IMPORTANT: Always look up current versions of actions before using them. You MUST NOT use action versions that don't exist.
Use the version lookup script to find the latest version and commit SHA for any action:
# Latest version + SHA for any action
scripts/action-versions <owner/repo>
# Specific tag
scripts/action-versions <owner/repo> --tag v4
scripts/action-versions <owner/repo>@v4
# All versions
scripts/action-versions <owner/repo> --all
Pin actions to the commit SHA with a version comment so Supply-chain attacks on the tag are mitigated:
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
For official actions/ actions, you can also cross-check at https://simonw.github.io/actions-latest/versions.txt
Prefer native builds to cross compiling. Github provides x86 and arm runners. It is better to run a build natively on a runner without emulation. Check references/runners.md for available runners.
For workflow_call (reusable workflows) and workflow_dispatch syntax, see references/reusable-workflows.md. Pay attention to input types, secrets: inherit, and how outputs flow between caller and callee.
MUST validate all workflow files with actionlint before considering the task done. Run actionlint on every workflow file that was created or modified. Fix all reported errors and warnings.
Always add concurrency groups to workflows triggered by pushes or PRs so new commits cancel in-progress runs for the same branch:
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.ref }}
cancel-in-progress: true
head_ref is the PR source branch name (e.g. feature-login); it is empty on push events, so the group falls back to github.ref. The github.workflow prefix prevents different workflows from colliding on the same branch. Always set cancel-in-progress: true explicitly — the default is false. For deploy workflows, omit cancel-in-progress or set it to false to avoid canceling a deploy mid-flight.
actions/checkout persists auth credentials in .git/config by default, allowing any subsequent step to push to the repo. Always set persist-credentials: false unless a later step explicitly needs to push:
- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
with:
persist-credentials: false