auto

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted user-supplied descriptions to determine execution paths.
  • Ingestion points: The primary ingestion point is the user-provided [plain-english goal] argument defined in SKILL.md.
  • Boundary markers: The skill incorporates several procedural boundaries, including a requirement for the agent to 'reason out loud' during classification and a 'Confirm' phase that mandates user approval before the selected route is executed.
  • Capability inventory: The skill possesses capabilities for tool invocation (Skill, Task), file reading (Read, Grep, Glob), and user interaction (AskUserQuestion), but does not have access to network or direct system-level execution tools.
  • Sanitization: No explicit sanitization or filtering of the goal string is performed; instead, the skill relies on the agent's internal classification logic and the provided routing-rules.md to map input to safe internal routes.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 01:09 AM
Security Audit — agent-trust-hub — auto