explore

Pass

Audited by Gen Agent Trust Hub on Feb 20, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill interpolates user-provided topics ($ARGUMENTS) and content from the local codebase directly into sub-agent prompts. * Ingestion points: User-provided query and codebase files read via Read or Grep tools. * Boundary markers: Rules such as exploration-agents.md include instructional constraints like 'Scope: ONLY read files directly relevant... Do NOT explore the entire codebase'. * Capability inventory: Access to Bash, Read, Grep, and TaskCreate tools. * Sanitization: None; untrusted content is interpolated directly into system instructions for sub-agents.
  • [Command Execution] (SAFE): The skill executes a bundled script 'scripts/dependency-mapper.sh' using the Bash tool. * Analysis: The script performs static analysis using standard shell utilities (find, grep, sort) to identify import patterns and coupling. It does not perform network operations or attempt privilege escalation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 20, 2026, 10:11 PM