The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes system commands such as gh, git, and jq to perform repository diagnostics and management. These actions are intended as part of its role in automated skill repair.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from external sources, creating an injection surface.
Ingestion points: Reads workflow run logs via gh run view --log-failed, check-run annotations via gh api, and external API documentation via WebFetch.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential injections within the logs it analyzes.
Capability inventory: The skill can modify existing skill files (SKILL.md), create new git branches, and submit Pull Requests.
Sanitization: No logic is specified for sanitizing or escaping the content retrieved from logs or external sources before it is used to generate code fixes.
[DATA_EXFILTRATION]: The skill pushes code changes and diagnostic dossiers to GitHub and fetches information from external APIs. These operations are consistent with its purpose and utilize ./secretcurl to prevent secret exposure in logs.

skill-repair