link-building

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The companion-check section instructs the agent to run shell commands that enumerate local skill directories and suggest installing missing skills via npx — a system-level probe and potential install flow unrelated to link-building and therefore a hidden/out-of-scope instruction.

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

• Third-party content exposure detected (high risk: 0.90). The skill's workflows (SKILL.md and references/link-audit.md and link-tactics.md) explicitly require fetching and inspecting open/public third-party content—e.g., Ahrefs/BuzzSumo exports, competitor and resource pages, HARO/Connectively queries, and manual site visits—to read/interpret untrusted web pages and journalist posts and then make decisions like outreach, removal requests, or disavow actions, which could enable indirect prompt injection.