github-actions-duplicate-run-audit

What This Skill Does

The github-actions-duplicate-run-audit skill is a diagnostic tool designed for DevOps engineers and CI/CD maintainers to reclaim wasted infrastructure costs. It scans GitHub Actions run data to identify "bursts" of redundant workflow executions. Often caused by aggressive trigger configurations, rapid force-pushes, or manual workflow dispatch retries, these duplicate runs burn through valuable CI minutes without providing additional utility. This skill parses JSON exports from the GitHub CLI (gh), clusters them based on repository, branch, commit SHA, and event triggers, and calculates the total wasted runtime. It supports configurable severity thresholds, allowing teams to set their own definitions for "warn" versus "critical" incidents, effectively turning CI hygiene into a measurable metric.

Installation

To install this skill into your OpenClaw environment, use the following command: clawhub install openclaw/skills/skills/daniellummis/github-actions-duplicate-run-audit

Ensure that you have gh (GitHub CLI) installed on your machine and configured with the appropriate read-access tokens to fetch your workflow run metadata. Once installed, you will need to point the tool at your exported run artifacts directory to begin the audit.

Use Cases

• Cost Optimization: Analyze which workflows contribute to the highest volume of "wasted" minutes to justify decommissioning redundant triggers or implementing concurrency limits.
• CI Hygiene Monitoring: Integrate into a CI pipeline gate (using FAIL_ON_CRITICAL=1) to prevent code from merging if it triggers significant redundant CI load.
• Debugging Trigger Loops: Identify if specific event combinations (e.g., push + pull_request) are causing double-execution storms during active development sprints.

Example Prompts

1. "Audit all CI runs in the artifacts/github-actions folder and show me a text summary of the most redundant workflows."
2. "Find any duplicate GitHub Action runs from the last week that wasted more than 60 minutes and report them in JSON format."
3. "Run a duplicate audit on my local fixtures, flagging any group with more than 3 repeats as a critical issue."

Tips & Limitations

• Data Dependency: This skill requires pre-fetched data. Ensure your collection script captures at least the databaseId, headSha, and startedAt fields to enable accurate grouping.
• Window Sensitivity: The DUPLICATE_WINDOW_MINUTES setting is critical. If your builds take longer than 30 minutes, consider increasing this window to avoid false positives where long-running builds are misidentified as duplicates.
• Scope: The skill performs best when filtering by specific repositories; running it across a massive enterprise-wide data dump may require adjusting the TOP_N limit to avoid overwhelming output.