SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsactrun-debug

actrun-debug

Testing & quality

View repository

Updated 10-15 21:02

About this skill

actrun-debug Skill

When an actrun workflow execution fails, identify the cause and propose a fix.

How to use this skill

Apply this skill when the user requests any of the following:

  • actrun execution failed and they want to know the cause
  • I want to debug workflow errors
  • they want actrun output analyzed

Diagnostic procedure

Step 1: Obtain the execution result

# 最新の実行一覧を確認
actrun run list
Obtain details of the failed run

actrun run view

Obtain detailed information in JSON (easier to analyze)

actrun run view --json


### Step 2: ログの確認

```bash
# 全ログを確認
actrun run logs <run-id>

# 失敗したタスクのログを特定して確認
actrun run logs <run-id> --task <job_id>/<step_id>

run view --json の steps From the array, status: "failed" identify the step of id を --task and pass it to

Step 3: Cause classification and remediation

Classify into the following patterns and remediate:

A. Tool not installed

Symptoms: command not found, No such file or directory

Remediation:

# 依存ツールチェック
actrun doctor

# ローカルツールがあるならスキップ設定
# actrun.toml:
#   local_skip_actions = ["actions/setup-node"]

B. Action compatibility issues

Symptoms: unsupported action input, action not found

Remediation:

  • Check whether it is a built-in supported action (checkout, cache, artifact, setup-node)
  • For unsupported actions, --skip-action skip with
  • Remote action is --trust allow fetching with the flag
actrun ci.yml --skip-action unsupported/action --trust

C. Path/workspace issues

Symptoms: file not found, path is different

Remediation:

  • --workspace-mode change (local → worktree or tmp）
  • actions/checkout if skipping, check whether the file is in the current directory
# worktree で隔離実行
actrun ci.yml --worktree

D. Secrets/variables not set

Symptoms: ${{ secrets.XXX }} is empty, authentication error

Remediation:

# シークレットを環境変数で渡す
ACTRUN_SECRET_GITHUB_TOKEN=$(gh auth token) actrun ci.yml

# .env ファイルから読み込み
actrun ci.yml --env .env.local

E. Conditional expression errors

Symptoms: if condition evaluation error, unexpected skip

Remediation:

  • --dry-run check the execution plan with
  • ACTRUN_LOCAL=true check the conditional branching by
  • ${{ env.ACTRUN_LOCAL }} check whether is set

F. Nix-related

Symptoms: nix develop error, package not found

Remediation:

# Nix を無効化して切り分け
actrun ci.yml --no-nix

# アドホックパッケージを指定
actrun ci.yml --nix-packages "nodejs python312"

G. Docker/container-related

Symptoms: docker: command not found, image pull failure

Remediation:

# Docker が動いているか確認
actrun doctor

# ランタイム変更
actrun ci.yml --container-runtime podman

Step 4: Retry

# 修正後、失敗ジョブのみリトライ
actrun ci.yml --retry

# 特定ジョブだけ実行
actrun ci.yml --job <failed-job-id>

JSON output analysis points

actrun run view <run-id> --json Important fields of :

Field Description
state completed, partial_failed, failed
ok Overall success/failure (boolean)
exit_code Exit code
steps[].id Step ID (job_id/step_name)
steps[].status success, failed, skipped
steps[].message Output message
tasks[].stdout_path stdout log file path
tasks[].stderr_path stderr log file path
issues[] Warnings/issues during execution

Reading log files directly

It is also possible to read the run store's log files directly:

# デフォルトの run store パス
_build/actrun/runs/<run-id>/tasks/<task_id>.stdout.log
_build/actrun/runs/<run-id>/tasks/<task_id>.stderr.log

Reference