puzzle-activity-planner
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user input (event themes, audience descriptions, and goal statements) to generate activity content and puzzle parameters.
- Ingestion points: User-provided event descriptions and themes described in the 'Process' section are interpolated into generated output and URLs.
- Boundary markers: The 'Execution boundaries' section provides explicit instructions to the agent to 'Never include student names, private records or sensitive classroom details' in URLs and suggests using local puzzle content if services are unavailable.
- Capability inventory: The skill does not have capabilities for file system access, subprocess execution, or direct network requests. It only generates text and links for the user's browser.
- Sanitization: The instructions explicitly direct the agent to 'Encode parameters with a URL library' and verify service compatibility.
- [NO_CODE]: The skill consists entirely of markdown instructions and metadata. It does not include any scripts (.py, .js, .sh), configuration files for package managers (package.json, requirements.txt), or compiled binaries.
Audit Metadata