pdd-prompting
Fail
Audited by Snyk on Nov 21, 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 legitimate PDD authoring guide with protective rules, but its Autopilot section lets the agent change approved rules and decide product-direction choices without asking the person, which is disclosed and listed yet may go beyond what the user expects.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (medium risk: 0.30). The Autopilot core job rules direct the agent to read GitHub issues as the job's input (triggering the PR, choosing a skill, and judging "vague" or "trusted" issues), and external users can post to those issues without the agent first selecting one, so this is medium.
MEDIUM W013: Attempt to modify system services in skill instructions.
- Attempt to modify system services in skill instructions detected (low risk: 0.10). Gate 1 passes only weakly: the skill documents starting a remote Autopilot core job (
pdd autopilot start --engine core) that creates PRs and changes issue labels, but gate 2 fails because no explicit irreversible operational loss is described.
Issues (3)
E004
CRITICALPrompt injection detected in skill instructions.
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
W013
MEDIUMAttempt to modify system services in skill instructions.
Audit Metadata