lean-canvas
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns detected. The skill follows established business strategy frameworks and uses local, verifiable scripts.\n- [COMMAND_EXECUTION]: The skill recommends running a local validation script
lean_canvas_validator.py. The script is included in the skill distribution and relies entirely on the Python standard library with no network or unauthorized system access.\n- [INDIRECT_PROMPT_INJECTION]: The skill parses user-supplied JSON data to validate business models. The attack surface is well-contained and poses minimal risk.\n - Ingestion points: Processes user-provided
canvas.jsonvia the validator script.\n - Boundary markers: The script enforces a structured JSON schema for input data.\n
- Capability inventory: The script is limited to local file reading and generating text-based validation reports.\n
- Sanitization: Uses standard
json.loadsand performs quality checks on text content without executing any part of the input.
Audit Metadata