verification-checkpoint

What This Skill Does

The verification-checkpoint skill acts as a critical quality assurance guard for OpenClaw AI agents. Its primary purpose is to enforce a mandatory verification phase before the agent claims that a task is complete. By integrating this skill, users ensure that no code or feature is marked as 'done' until it has successfully passed local unit tests, type checks, build processes, and manual validation. The skill provides a structured framework for validating technical output, effectively preventing common pitfalls such as deploying broken code, ignoring compilation errors, or making untested assumptions. It supports varying levels of verification—from minimal checks for typo corrections to full-scale validation for critical architectural changes.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/sa9saq/verification-checkpoint Ensure your project repository has the appropriate testing frameworks (e.g., pytest, jest) and build tools (e.g., npm, tsc) configured, as the skill relies on these native CLI commands to perform its validation cycles.

Use Cases

• Continuous Delivery Flows: Use this to prevent PRs from being generated with failing test suites.
• Refactoring Safety: Ensures that large-scale code changes do not break existing build pipelines.
• Automated Bug Resolution: Validates that a bug fix actually resolves the issue by running specific tests before notifying the user.
• Strict Quality Control: Enforces mandatory type checking in TypeScript environments to maintain high code stability.

Example Prompts

1. "I've finished refactoring the authentication module. Please run the verification-checkpoint and report back."
2. "Implement the new API endpoint and ensure the verification-checkpoint passes Level 3 before confirming completion."
3. "Please fix the linting errors in the user profile component, then verify the build and run the full test suite."

Tips & Limitations

To maximize the utility of this skill, ensure that your package.json or pyproject.toml scripts are clearly defined, as the agent relies on these to execute commands correctly. The skill works best when paired with an automated test suite. Note that while it can execute commands, it cannot fix architectural design flaws. If the verification fails, the agent will enter an error-handling loop; providing descriptive error messages to the agent will significantly improve its success rate in applying automated fixes.