anysearch

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md, README, and CLI scripts direct the agent to call the AnySearch API (POST to https://api.anysearch.com/mcp) for web searches and to run the "extract" command on arbitrary URLs and to call "list_domains" and "read query_format" from API responses — meaning the agent ingests untrusted public web/UGC content (web pages, search results, domain listings) which it must interpret and use to construct subsequent searches or actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime CLI scripts call the external JSON-RPC endpoint https://api.anysearch.com/mcp to fetch list_domains and other tool responses that the agent is required to follow (e.g., obey returned query_format and params_schema), so remote content directly controls agent prompts/behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The prompt explicitly instructs invoking PowerShell with "-ExecutionPolicy Bypass" (bypassing a security restriction) and directs filesystem writes of API keys (even though with user consent), so it asks the agent to bypass a security mechanism and modify on-disk state.