debug

Pass

Audited by Gen Agent Trust Hub on Apr 13, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool for investigative purposes, specifically running git diff to see recent changes and executing tests to reproduce errors. These operations are standard for development and debugging workflows.- [PROMPT_INJECTION]: The skill processes untrusted data such as error messages and stack traces, which constitutes an indirect prompt injection surface. Ingestion points: Input data from error reports, logs, and stack traces. Boundary markers: Absent. Capability inventory: Access to Bash, Read, Grep, and Glob tools for file system interaction and command execution. Sanitization: No explicit sanitization or use of delimiters for external content is defined. This is a known surface for debugging tools and is managed by the agent's internal safety guardrails.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 13, 2026, 04:17 AM
Security Audit — agent-trust-hub — debug