The Agent Skills Directory

[SAFE]: The skill instructions strictly maintain a read-only posture on the target repository and do not include any scripts, external downloads, or persistence mechanisms.
[COMMAND_EXECUTION]: The skill directs the agent to run the repository's own verification and version-control tools (such as test runners, linters, or git) to gather evidence for the audit. This is consistent with its stated purpose of establishing a quality baseline.
[PROMPT_INJECTION]: As an audit tool that processes repository content, the skill has an inherent surface for indirect prompt injection. 1. Ingestion points: The agent reads source code, configuration files, and CI/CD workflows from the repository. 2. Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions within the ingested repository data. 3. Capability inventory: The agent is authorized to execute local verification commands as part of the audit process. 4. Sanitization: The skill does not describe specific sanitization or filtering steps for the untrusted repository content before processing.

quality-baseline