The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a bundled Python script scripts/inspect_pr_checks.py that executes system commands to interact with the GitHub environment. It uses subprocess.run with list-based arguments to prevent shell injection vulnerabilities while calling the git and gh (GitHub CLI) binaries.
Evidence: Functions run_gh_command and find_git_root in scripts/inspect_pr_checks.py use subprocess.run to call gh auth status, gh pr view, gh pr checks, gh run view, and git rev-parse.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface because it ingests external, untrusted content from GitHub Actions logs and passes it to the agent for summarization and fix planning.
Ingestion points: scripts/inspect_pr_checks.py fetches logs from GitHub Actions runs and jobs via the gh CLI.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used when providing log snippets to the agent.
Capability inventory: The skill is granted Write, Edit, and Bash capabilities, which are high-impact tools used to implement the fix plan generated from the logs.
Sanitization: Log content is extracted based on failure markers (e.g., 'error', 'fail', 'traceback') but is not sanitized for natural language instructions that might trick the agent into performing unintended actions.

openai-gh-fix-ci