copying-endpoints-across-projects

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

• Third-party content exposure detected (medium risk: 0.65). SKILL.md step 2/6 requires reading and re-sending the source endpoint’s full query via endpoint-get and then passing that captured text verbatim into endpoint-create (outsider-authored endpoint query/config could be attacker-controlled free text that enters the agent’s LLM context during summarization and prompt construction).