zig-http
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external network sources through HTTP server and client implementations.
- Ingestion points: The server implementation in
SKILL.mdreads request bodies viareq.readAlland inspects request targets (URLs) for routing logic. - Boundary markers: The instructions do not provide delimiters or warnings to treat external network content as untrusted data, potentially leading the agent to follow instructions embedded in HTTP responses or request bodies.
- Capability inventory: The skill enables full network capabilities, including making arbitrary HTTP requests, uploading files via multipart forms, and managing WebSocket connections.
- Sanitization: There is no evidence of input validation, escaping, or sanitization of the data received from network operations before it is used in program logic or displayed.
Audit Metadata