solid-generic
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill uses the fuse-ai-pilot:explore-codebase agent to ingest and analyze project codebases, which constitutes an untrusted data source. The agent possesses capabilities to modify the file system (Bun.write, unlink) and make network requests (fetch) as shown in the provided templates. There are no explicit boundary markers or sanitization requirements defined in the instructions to prevent the agent from following malicious commands embedded in the codebase.
- Ingestion points: Project source files and directory structure.
- Capability inventory: File system access via Bun and Node.js APIs, and network access via the fetch API.
- Sanitization: The instructions lack specific guidance on escaping or validating data processed from the codebase before use in prompt interpolation or execution.
Audit Metadata