docs-gen
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes library source code and documentation to generate content for other AI agents. This ingestion of untrusted data from the repository (source code, tests, and documentation) creates an attack surface for indirect prompt injection. The skill lacks sanitization of ingested content and does not provide boundary markers in its output to prevent consumer AIs from executing malicious instructions embedded in the documentation.
- [COMMAND_EXECUTION]: The skill executes various shell commands, including node, git, and ripgrep (rg), to facilitate documentation building and synchronization.
- [DYNAMIC_EXECUTION]: The skill deploys and executes a bundled Node.js script (push-docs.mjs) within the workspace to handle documentation uploads.
- [DATA_EXFILTRATION]: The skill transmits repository documentation and metadata to a remote server specified by the user-defined DOCS_SERVER_URL environment variable.
Audit Metadata