The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is primarily focused on processing 'External Reviewer' feedback, which represents a potential vector for indirect prompt injection if malicious instructions are embedded in the review comments.
Ingestion points: External code review feedback and GitHub PR comments (as described in SKILL.md).
Boundary markers: Absent. The skill does not define specific tags or delimiters to isolate untrusted feedback from the agent's internal instructions.
Capability inventory: The skill instructs the agent to use grep for codebase searching and gh api for replying to GitHub comments. It also implies file modification capabilities for implementing fixes.
Sanitization: Absent. There are no instructions for escaping or filtering the content of the external feedback, although the skill explicitly instructs the agent to 'Verify before implementing' and to 'Push back with technical reasoning if wrong', which serves as a logical defense-in-depth measure.
[COMMAND_EXECUTION]: The skill provides specific instructions on using command-line tools to fulfill its tasks.
Evidence: The skill references the use of grep to verify code usage (YAGNI checks) and gh api repos/{owner}/{repo}/pulls/{pr}/comments/{id}/replies to communicate on GitHub. These are standard developer tools, but they represent the skill's operational capabilities.

receiving-code-review