The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes code review feedback from external sources, which constitutes an ingestion point for untrusted data that could contain malicious instructions.
Ingestion points: External reviewer feedback is ingested and evaluated as described in references/code-review-reception.md.
Boundary markers: The instructions do not define strict technical delimiters for the external feedback strings, although they advise restating requirements for clarity.
Capability inventory: The skill can execute shell commands like git and run project tests via scripts/find-polluter.sh.
Sanitization: The skill provides strong instructional mitigations, requiring the agent to be skeptical of external feedback and to perform independent technical verification before implementing any suggested changes.
[COMMAND_EXECUTION]: The skill includes a utility script scripts/find-polluter.sh for debugging test suite pollution.
Evidence: The script executes npm test "$TEST_FILE" for files matching a provided pattern. This is a functional requirement for identifying state pollution in the development environment.

code-quality