The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of CI logs fetched from GitHub Actions.\n
Ingestion points: scripts/parse_check_failures.py retrieves logs via the gh run view command.\n
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the log data when proposing fixes.\n
Capability inventory: The skill uses the Bash tool for git operations and the Edit tool to modify source code, which could be exploited if malicious logs influence the agent's decisions.\n
Sanitization: There is no sanitization of the log content; it is parsed using regex patterns that could be bypassed or manipulated.\n- [COMMAND_EXECUTION]: The helper scripts scripts/check_pr_status.py and scripts/parse_check_failures.py use subprocess.run to execute GitHub CLI (gh) commands. While these calls use argument lists to avoid shell injection, they represent a surface for interacting with system utilities based on external identifiers.\n- [EXTERNAL_DOWNLOADS]: The skill fetches CI logs and metadata from GitHub's official infrastructure using the gh command-line tool to facilitate the fix loop.

sc-pr-fix