The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using gh (GitHub CLI) and git to retrieve pull request metadata and perform repository operations.
Evidence: gh pr view --json baseRefName --jq '.baseRefName' and git fetch origin <base-branch> are used to resolve the merge target.
Evidence: git diff origin/<base-branch>...HEAD is used to generate the code diff for review.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data that is later used to drive implementation decisions.
Ingestion points: Untrusted data enters the context through PR comments fetched in Step 1 and the code diff generated in Step 3.
Boundary markers: The instructions do not define clear delimiters or include warnings for the agent to ignore potentially malicious instructions embedded within the PR comments or the code under review.
Capability inventory: The skill possesses the capability to modify the codebase through the /resolve-findings skill (Step 5), which acts upon the findings identified in previous steps.
Sanitization: There is no mention of sanitizing, escaping, or filtering the external content before it is passed to the evaluation and resolution phases.

review-pr