The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local repository (source code, git history, and documentation) which could contain adversarial instructions.
Ingestion points: The skill specifies commands like cat, git log, and rg (ripgrep) to read local file content into the agent's context for observation and interpretation.
Boundary markers: No boundary markers or "ignore instructions" warnings are implemented for the ingested data to prevent the agent from following instructions found within the code.
Capability inventory: The skill utilizes subprocess calls for git, rg, and npx (specifically executing tsc, eslint, and vitest).
Sanitization: No sanitization or validation of the ingested content is performed before the agent processes it.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to perform repository analysis and validation.
Uses rg (ripgrep) to search for specific patterns, such as lowercase component exports or server actions, in TypeScript and TSX files.
Uses git log and git blame to inspect commit history and verify if patterns are intentional based on authorship.
Uses npx to run validation tools including the TypeScript compiler (tsc), eslint, and vitest. While these are standard developer tools, npx can be used to execute packages from the local environment.

audit-code-quality