Identify and rank flaky GitHub Actions steps by analyzing success and failure outcomes across multiple workflow runs.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install github-actions-step-flake-audit
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install github-actions-step-flake-audit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is GitHub Actions Step Flake Audit?
The GitHub Actions Step Flake Audit is a specialized tool designed to solve the persistent problem of intermittent CI failures. By integrating this capability into your workflow using Openclaw Skills, developers can automatically scan exported GitHub Actions run data to find job steps that alternate between passing and failing. This skill provides a quantitative approach to managing technical debt, allowing engineering teams to prioritize fixing the most disruptive flakes based on failure rates and volume.
Unlike simple pass/fail logs, this skill aggregates data across many runs to highlight patterns of instability. Using Openclaw Skills for CI auditing ensures that your development process remains lean by identifying non-deterministic steps that would otherwise go unnoticed until they cause a major release delay.
GitHub Actions Step Flake Audit Use Cases
- Identifying non-deterministic test steps in complex CI/CD pipelines.
- Auditing repository health to reduce re-run culture among developers.
- Enforcing quality gates in CI by failing builds when critical flake thresholds are met.
- Benchmarking pipeline stability across different branches or repositories.
How GitHub Actions Step Flake Audit Works
- Collect GitHub Actions run data in JSON format using the GitHub CLI (gh).
- The skill parses these JSON files based on a configurable glob pattern to extract job and step conclusions.
- Outcomes are grouped by repository, workflow, job, and specific step name to establish a baseline of behavior.
- A flake score is calculated for each group by comparing the ratio of successes to failures.
- The skill generates a ranked report in text or JSON format, highlighting the most unstable steps based on user-defined thresholds.
GitHub Actions Step Flake Audit Setup
Ensure you have bash and python3 installed. You will also need the GitHub CLI (gh) to export run data. To set up this Openclaw Skills instance:
- Export your run history to a local directory:
gh run view <run-id> --json databaseId,workflowName,headBranch,headSha,url,repository,jobs > artifacts/github-actions/run-<run-id>.json
- Execute the audit script with your desired configuration:
RUN_GLOB='artifacts/github-actions/*.json' \
MIN_OCCURRENCES=5 \
bash skills/github-actions-step-flake-audit/scripts/step-flake-audit.sh
GitHub Actions Step Flake Audit Data Schema & Taxonomy
The skill processes specific metadata fields from GitHub Actions JSON exports to build a flake profile. It looks for the following structure:
| Field | Description |
|---|---|
| databaseId | Unique ID of the workflow run |
| workflowName | Name of the CI workflow |
| jobs | Array containing job names and step conclusions (success/failure) |
| repository | The source repository for the run data |
When using Openclaw Skills for reporting, the JSON output mode will provide a summary object including ranked groups and a list of specific critical groups that exceeded your failure thresholds.
GitHub Actions Step Flake Audit Advanced Features
- Customizable failure rate thresholds using WARN_FAILURE_RATE and CRITICAL_FAILURE_RATE for fine-grained alerting.
- Regex-based filtering with REPO_MATCH and WORKFLOW_EXCLUDE to target specific areas of your codebase.
- Fail-gate enforcement via the FAIL_ON_CRITICAL setting, allowing Openclaw Skills to automatically block builds if flakes exceed limits.
- Support for large-scale audits across thousands of JSON files using the RUN_GLOB pattern.
SKILL.md
Loading
Related Openclaw Skills
GitHub Actions SHA Rerun Debt Audit
Audit GitHub Actions rerun debt to identify and eliminate commit-level CI inefficiencies.
GitHub Actions Self-Hosted Risk Audit
A specialized security auditing tool for identifying high-risk configurations in GitHub Actions workflows utilizing self-hosted runners.
GitHub Actions Secret Exposure Audit
An automated auditing tool to identify and mitigate secret exposure risks within GitHub Actions workflow configurations.
GitHub Actions Runtime Regression Audit
Analyze and compare GitHub Actions runtime metrics to detect performance regressions and optimize CI costs.
GitHub Actions Stuck Run Audit
A monitoring skill designed to detect and alert on GitHub Actions workflows that remain in queued or in-progress states for excessive periods.
GitHub Actions Timeout Risk Audit
Proactively identify and fix near-timeout GitHub Actions jobs before they break your CI pipeline.
METADATA
Featured*
