The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local shell script scripts/poll_pr_status.sh to retrieve the current state of a GitHub Pull Request. This script uses the gh CLI to interact with the GitHub API.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its automated processing of untrusted external data.
Ingestion points: The scripts/poll_pr_status.sh script ingests CI check names and PR comment bodies (specifically targeting CodeRabbit bot comments) using gh pr checks and gh api respectively.
Boundary markers: The instructions lack explicit boundary markers or delimiters to isolate the ingested external text from the agent's internal instructions.
Capability inventory: The skill possesses significant capabilities, as it is designed to autonomously invoke other skills like /fix-ci and /fix-review which modify the codebase and push changes back to the repository.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from GitHub comments or CI check results before they influence the agent's decision-making loop.
[AUTONOMOUS_BEHAVIOR]: The skill is designed to operate with high autonomy, looping until CI passes and proactively triggering itself via PostToolUse hooks without direct user confirmation for each iteration.

pr-guardian