find-agent-skills

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

• Third-party content exposure detected (high risk: 0.85). The workflow requires opening and inspecting outsider-authored skillZs detail pages and their SKILL.md/descriptions from https://skillzs.dev/... at runtime, which are public web content that can contain free text (including prompt-injection payloads) read into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill workflow explicitly instructs the agent at runtime to query and consume results from https://skillzs.dev/browse?q=, meaning external content fetched from that URL will directly influence the agent's prompts and decisions.