design

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt requires "complete code (no ellipsis)" and "exact commands" plus exhaustive, copy‑pasteable steps after inspecting the codebase, which can force the agent to reproduce any embedded secrets (API keys, tokens, .env values) verbatim in its outputs.

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

• Third-party content exposure detected (high risk: 0.90). The skill's Research Phase and Research Tools (references/research-phase.md and the "Research Tools" list) explicitly fetch and ingest open-web/library content via mcp__perplexity__perplexity_research, mcp__exa__web_search_exa, and mcp__context7__query-docs and then feed those findings into the plan and review agents, so untrusted third‑party content can influence decisions and agent actions.