fusion-discover-skills

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly falls back to reading public GitHub repositories and using gh/GraphQL/code search to read SKILL.md and repository content (agents/github-mcp-advisor.md and agents/github-raw-search-advisor.md), so it ingests untrusted, user-generated third-party content from GitHub that can influence discovery and lifecycle actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs read-only runtime fetching of SKILL.md from the GitHub catalog (e.g., https://github.com/equinor/fusion-skills via gh api/graphql or gh search) and to inject that remote content into the agent's context to derive recommendations, meaning external content fetched at runtime can directly control prompts/instructions.