project-ops-router

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed as a routing layer. It identifies task types (Git operations, server coordination, project memory, etc.) and hands them off to specialized agents, which is a security best practice for compartmentalization.
  • [COMMAND_EXECUTION]: The skill uses git rev-parse --git-common-dir and --show-toplevel via the Bash tool. These commands are used strictly for environment detection (scope identification) and do not involve state modification or external network access.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads memory/BRIEFING.md. This is a local read operation within the project workspace used to determine the current project phase and active worktrees for routing context.
  • [PROMPT_INJECTION]: No evidence of role-play overrides, system prompt extraction, or safety filter bypass instructions were found in the skill body or references.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from memory/BRIEFING.md. While this constitutes an attack surface where untrusted data could influence routing, the risk is minimal as the ingested content is used solely to select a child skill and the router is explicitly forbidden from executing complex logic or connecting to remote servers.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 03:15 AM
Security Audit — agent-trust-hub — project-ops-router