Skills
skills/halt-catch-fire/skills/web-search/Gen Agent Trust Hub

web-search

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized data access detected.
  • [EXTERNAL_DOWNLOADS]: Mentions installing the belt CLI via npx and references documentation from the official inference-sh GitHub repository. These are legitimate resources associated with the skill's primary purpose.
  • [COMMAND_EXECUTION]: Utilizes the belt CLI for interacting with search and extraction apps. Command execution is appropriately restricted in the skill configuration to the belt tool suite.
  • [DATA_EXPOSURE]: Demonstrates workflows where external web content is ingested and processed by an LLM. The skill uses boundary markers (e.g., <search-results> tags) to help delimit untrusted data, reducing the risk of indirect prompt injection during RAG operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 02:15 AM
Security Audit — agent-trust-hub — web-search