The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates by scanning and extracting content from a target codebase to generate an audit report. This creates a surface for indirect prompt injection if the analyzed code contains malicious instructions intended to mislead the agent or influence its output.
Ingestion points: Grep, glob, and file read operations across the target project directory.
Boundary markers: The report template uses Markdown code blocks to enclose findings, which provides structural separation but lacks explicit instructions to the agent to disregard instructions embedded within code snippets.
Capability inventory: The skill utilizes file system read access, grep/glob search, and invocation of external linters (tsc, depcheck) via subprocess execution.
Sanitization: There is no evidence of sanitization or escaping of the content read from files before it is interpolated into the final report.
[SAFE]: The skill uses industry-standard development and static analysis tools including npx tsc, npx depcheck, and npm audit. These tools are invoked correctly for their intended purposes (checking for unused dependencies, compilation errors, and known vulnerabilities) and do not involve untrusted remote code execution.

code-review