find-skills

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests public, user-generated marketplace catalogs and SKILL.md files from external sites (e.g., Phase 3's WebFetch of https://raw.githubusercontent.com/dan323/easier-life-skills/... and https://raw.githubusercontent.com/anthropics/skills/..., and Phase 4's WebSearch/scan of GitHub repos to extract SKILL.md), and it uses those third-party contents to decide which skills to recommend, so untrusted content could influence agent actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches marketplace catalogs at runtime from https://raw.githubusercontent.com/dan323/easier-life-skills/master/.claude-plugin/marketplace.json and https://raw.githubusercontent.com/anthropics/skills/master/.claude-plugin/marketplace.json using WebFetch, and that remote JSON is injected into the agent's candidate pool and used to drive recommendations (i.e., directly influences agent prompts/context), so these URLs are runtime external dependencies that control the agent.