atmee-setup
Fail
Audited by Snyk on Nov 19, 2026
Risk Level: CRITICAL
Full Analysis
CRITICAL E004: Prompt injection detected in skill instructions.
- Potential prompt injection detected (medium risk: 0.30). The skill's purpose is setting up the atmee server and checks, and its risky steps (adding the MCP server, the Codex network-access change, and suppressing tool prompts) are disclosed and user-approved, but the permanent suppression of Claude Code's per-call tool prompts weakens a protection beyond what the setup needs, so I flag it as medium under recall-first.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (medium risk: 0.65). The skill has the agent edit persistent agent config outside the project (e.g.
~/.claude.json,~/.codex/config.tomlto enable sandbox network access for every session,~/.claude/settings.jsonto pre-approve tool permissions), which weakens the agent's security controls, though it asks the creator first and sudo is not involved.
Issues (2)
E004
CRITICALPrompt injection detected in skill instructions.
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata