skill-maker

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md Step 2 ("Research before writing") explicitly requires using WebSearch/WebFetch to retrieve and read web pages and GitHub/repos and to convert that external content into skill files, so the agent will fetch and interpret untrusted public third‑party content as part of its workflow.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's Step 2 requires using WebSearch/WebFetch to retrieve external web pages at runtime and to "preserve 100% of the data" when converting fetched pages into skill content, so runtime-fetched external URLs (i.e., arbitrary web pages retrieved via WebFetch) can directly control prompts/instructions — flagged: WebFetch / external web pages (runtime-fetched URLs).