review-roadmap

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill requires emitting findings that include "a short quote" from the roadmap and writing a findings bundle, which forces the LLM to reproduce file contents verbatim (risking secret exfiltration if the roadmap contains secrets).

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

• Third-party content exposure detected (high risk: 0.70). The required workflow reads the thread-root roadmap.md end-to-end at runtime (SKILL.md:11), and that roadmap.md is outsider-authored content unless the operating user authored it, creating an indirect prompt-injection risk via LLM ingestion of free text from that file.