mdbook-user-docs
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run various shell commands, including 'mdbook build', 'mdbook test', and repository-specific scripts for linting and style validation. Executing arbitrary scripts from the repository under analysis could be dangerous if the repository contains malicious shell code.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the project repository (such as documentation chapters, configuration files, and repository-level instructions) and has the capability to write files and execute commands. This creates an attack surface where instructions embedded in the project files could maliciously influence the agent's behavior.
- Ingestion points: Documentation chapters, 'book.toml', 'SUMMARY.md', and repository-level instruction files.
- Boundary markers: The instructions explicitly direct the agent to use 'safe isolation' or 'non-mutating checks' and to verify external scripts for absolute paths or side effects before execution.
- Capability inventory: File system modifications in 'Edit' mode and execution of system commands for building and testing.
- Sanitization: The agent is told to perform 'static review' instead of running commands if safe isolation is not possible.
Audit Metadata