creator-intelligence
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from local project files to build persistent context.
- Ingestion points: Reads CLAUDE.md files from home and project directories, and scans project files like package.json for auto-activation.
- Boundary markers: Does not utilize specific delimiters or escape sequences to separate data from instructions within the CLAUDE.md files.
- Capability inventory: Includes filesystem read/write, database operations (query, insert, update), browser automation, and email functionality.
- Sanitization: The skill instructions do not specify any sanitization or filtering of the content ingested from the context files.
- [COMMAND_EXECUTION]: The skill configuration defines extensive tool bindings for local operations.
- Evidence: The skill.yaml file explicitly maps MCP servers to capabilities such as reading/writing project files, querying/inserting into databases, and automating browser/email tasks.
Audit Metadata