how-it-works
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a bundled Python script
scripts/update_project_instructions.pyto automate the configuration of project-level instructions. This involves modifyingAGENTS.mdandCLAUDE.mdfiles in the user's project directory. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from project-specific documentation and instruction files (
AGENTS.md,CLAUDE.md, andHOW_IT_WORKS.md). This creates a vulnerability surface where instructions hidden in these files could influence the agent's behavior during the documentation process. - Ingestion points: Reads project-level instruction files (
AGENTS.md,CLAUDE.md) and existing project documentation. - Boundary markers: While the skill provides a framework for "source authority," it lacks specific delimiters to prevent the agent from accidentally following instructions contained within the ingested project files.
- Capability inventory: Executes Python scripts, performs file system writes, and conducts local tests.
- Sanitization: The workflow includes specific requirements to redact sensitive information and use synthetic data for examples.
Audit Metadata