coordinator

Fail

Audited by Snyk on Nov 16, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.30). The skill is a disclosed coordination workflow, but its command at lines 264-266 strips disable-model-invocation from third-party skills so the agent can invoke them autonomously. That weakens an upstream protection beyond what the routing needs, though it is openly documented and not aimed at deceiving or attacking the user.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (medium risk: 0.30). The required gate reads GitHub PR diffs and metadata on every PR, and the babysit and triage routes read PR review comments and raw issues that outsiders can post on, all without the agent first selecting a specific item, so this is medium exposure.

MEDIUM W013: Attempt to modify system services in skill instructions.

  • Attempt to modify system services in skill instructions detected (medium risk: 0.40). The skill is a multi-agent coordination workflow, but it includes a shell command that uses sed -i to strip disable-model-invocation guards from third-party skill files in ~/.agents/skills, which is a modification of machine state that bypasses an invocation control; it also directs killing processes, though it requires no sudo, system files, accounts, or SSH changes.

Issues (3)

E004
CRITICAL

Prompt injection detected in skill instructions.

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

W013
MEDIUM

Attempt to modify system services in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 16, 2026, 02:27 AM
Issues
3