The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and mitigates the risk of processing untrusted data from the repository.
Ingestion points: Processes test files, comments, fixtures, and git history during the audit process, as outlined in SKILL.md and references/lane-auditor.md.
Boundary markers: Explicitly mandates that test data must never be interpreted as instructions and provides a dedicated security policy in shared/external-data.md to enforce this boundary.
Capability inventory: Capabilities include reading and writing files in the repository, running the project's test suite, and executing a local Node.js script for report generation.
Sanitization: Includes comprehensive guidelines in shared/external-data.md for safe shell command execution, avoiding interpolation of untrusted prose, and validating inputs using character allowlists.
[COMMAND_EXECUTION]: The skill uses local command execution to perform its primary functions.
Evidence: Instructs the agent to run the project's test suite for baselining and to execute the provided scripts/render-report.mjs script via node to generate final reports. The Node.js script includes security checks, such as refusing to write through symbolic links.

auditing-tests