The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its processing of external log data.
Ingestion points: The script scripts/inspect_pr_checks.py (lines 217 and 228) retrieves GitHub Actions logs using gh run view --log and the GitHub API.
Boundary markers: There are no instructions in SKILL.md or the script to use delimiters or ignore instructions embedded within the processed logs.
Capability inventory: The skill has the capability to execute gh and git commands and is explicitly instructed in SKILL.md (Step 7) to implement plans by committing and pushing code changes to the repository.
Sanitization: Log content is analyzed and summarized for the agent without filtering or sanitization of potentially malicious instructions.
[COMMAND_EXECUTION]: The script scripts/inspect_pr_checks.py executes system commands to interact with the repository host.
Evidence: The script uses subprocess.run (lines 53, 62, 114) to invoke the gh and git CLI tools. While the implementation follows best practices by using list-based arguments to prevent shell injection, the commands use inputs derived from user arguments and external PR metadata.

fix-github-ci