The Agent Skills Directory

[COMMAND_EXECUTION]: The bundled script scripts/inspect_pr_checks.py executes system commands via subprocess.run to interact with standard development tools.
It calls git rev-parse --show-toplevel to determine the project root.
It repeatedly invokes the GitHub CLI (gh) for tasks such as verifying authentication status (gh auth status), viewing pull request metadata (gh pr view), listing status checks (gh pr checks), and retrieving workflow logs (gh run view, gh api).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, creating a potential surface for indirect prompt injection attacks.
Ingestion points: Failure logs and snippets are retrieved from GitHub Actions via the gh CLI in scripts/inspect_pr_checks.py and presented to the agent for analysis.
Boundary markers: The instructions do not specify strict boundary markers or delimiting for the ingested log content.
Capability inventory: The skill is designed to draft and implement code changes based on these logs, and it has the ability to execute shell commands (git, gh) and perform network operations via the GitHub CLI.
Sanitization: There is no evidence of sanitization or filtering of the log content to remove potential embedded instructions before the agent processes it to create a fix plan.

gh-fix-ci