agent-deepweb

Pass

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection attack surface identified. The skill instructs the agent to process data fetched from external URLs via the 'agent-deepweb' tool, which could contain malicious instructions.
  • Ingestion points: Data returned from agent-deepweb fetch, graphql, and template tools.
  • Boundary markers: None explicitly defined in the tool output description or agent instructions to separate user data from instructions.
  • Capability inventory: Network requests (agent-deepweb), file system access (Read, Grep, Glob).
  • Sanitization: No sanitization of external content is specified before processing or interpolating responses into the agent's context.
  • [COMMAND_EXECUTION]: The skill documents usage of the 'agent-deepweb' CLI tool for managing authenticated HTTP requests and provides examples for shell commands including Bash, Read, Grep, and Glob.
  • [EXTERNAL_DOWNLOADS]: References the 'agent-deepweb' CLI tool hosted on the author's repository (github.com/shhac/agent-deepweb).
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 16, 2026, 01:20 PM
Security Audit — agent-trust-hub — agent-deepweb