The Agent Skills Directory

[SAFE]: The skill's primary purpose is to perform read-only auditing of GitHub workflow files to ensure Actions are pinned to secure hashes.
[SAFE]: It contains explicit restrictions prohibiting file modifications and mutating API calls (e.g., POST, DELETE), using only GET requests through the GitHub CLI.
[SAFE]: All external communications are conducted via the official GitHub CLI targeting a well-known service (github.com).
[PROMPT_INJECTION]: The skill processes untrusted data (content from GitHub workflow files) which presents a surface for indirect prompt injection.
Ingestion points: Workflow references are read from repo files in Step 2.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the parsed content.
Capability inventory: The skill uses Bash and gh tools for search and API interaction.
Sanitization: No specific sanitization of fetched Action references is defined before presenting them to the user. The risk is mitigated by the tool's limited reporting scope and human-in-the-loop SHA verification.

action-audit