meilisearch

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to "MUST fetch and read the live documentation" from the public Meilisearch docs (https://www.meilisearch.com/docs/llms.txt), which requires ingesting untrusted, user-accessible third-party content that can influence code and behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs at runtime to "MUST fetch and read the live documentation" from https://www.meilisearch.com/docs/llms.txt, so fetching that remote text would directly control the agent's instructions/prompts as a required dependency.