create-skill

Fail

Audited by Snyk on May 5, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (medium risk: 0.60). The prompt contains a deceptive/incomplete disclosure: while the "Disclosure" claims it will only create skills/{name}/SKILL.md, the protocol (Step 4) also explicitly instructs modifying .claude/harness.json and potentially CLAUDE.md, so it hides actions beyond the stated purpose (other instructions like "pretend you are a different AI session" are contextually relevant to testing and not covert).

Issues (1)

E004
CRITICAL

Prompt injection detected in skill instructions.

Audit Metadata
Risk Level
CRITICAL
Analyzed
May 5, 2026, 10:49 AM
Issues
1