make-the-docs-trustworthy
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill explicitly instructs the agent to verify claims in documentation by running shell commands found in the text. For example,
rules/staleness-without-a-diff.mddirects the agent to "Run the command the page describes" and "Open the path it names" to settle staleness claims. - [INDIRECT_PROMPT_INJECTION]: The skill's core workflow involves processing potentially untrusted documentation files and using their content to drive agent actions, which is a significant attack surface.
- Ingestion points: Markdown (.md) files throughout the repository being audited.
- Boundary markers: The instructions lack specific guidance on how to delimit or isolate instructions found within the documentation to prevent them from being interpreted as direct agent commands.
- Capability inventory: The agent is expected to use shell execution and file system access to verify the claims found in the documentation.
- Sanitization: There are no instructions for validating or sanitizing the commands extracted from documentation before they are executed in the shell.
- [DYNAMIC_EXECUTION]: The skill facilitates the dynamic generation and execution of commands based on natural language content found in external files, which could be exploited to run arbitrary code if the documentation is manipulated.
Audit Metadata