Ln 622 Build Auditor

What it does

The Ln 622 Build Auditor skill is a specialized AI agent worker designed to analyze build processes and code quality. It identifies potential inefficiencies, errors (like compiler/linter issues or failed tests), and configuration problems within a codebase. The tool then provides structured findings with severity levels, suggested solutions, and calculates a compliance score out of 10 for the "Build Health" category. It operates as part of an audit pipeline and requires specific context data to function correctly.

When to use it

• To proactively identify build health issues before they impact deployments.
• During code reviews to assess adherence to best practices and coding standards.
• As part of a continuous integration/continuous delivery (CI/CD) pipeline for automated code quality checks.
• When needing an assessment of compliance with defined "best practices" and principles.

Key capabilities

• Build Health Auditing: Checks compiler errors, linter warnings, deprecation notices, type errors, and test failures.
• Configuration Analysis: Evaluates build configurations to determine if they are development-only or production-ready.
• Severity Assessment: Categorizes findings by severity level.
• Effort Estimation: Provides estimates of the effort required to address identified issues.
• Recommendation Generation: Suggests specific actions for improving code quality and build health.
• Compliance Scoring: Calculates a "Build Health" compliance score (X/10).
• Trend Tracking: Logs build health metrics over time.

Example prompts

• “Analyze the codebase at /path/to/codebase for build health issues.”
• “Assess the compliance of this code with our defined best practices, focusing on potential deprecation warnings.”
• "Run a full build audit and generate a report in /output/directory/ln-622--global.md."

Tips & gotchas

• Mandatory Context: The skill requires specific context data including the tech stack (build tool, test framework), best practices, codebase root, and output directory to be provided via contextStore. Ensure this information is available.
• Reference Files: The skill relies on several reference files (audit_worker_core_contract.md, two_layer_detection.md, results_log_pattern.md, and a report template). These must be accessible to the agent.
• Layered Detection: The tool uses a two-layered detection methodology, first running basic checks then analyzing output for more nuanced issues like deprecation warnings.