The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a surface for processing external content by instructing the agent to fetch documentation from https://cursor.com/docs/api/origin/* at runtime. This ingestion point is mitigated by clear boundary instructions that forbid the agent from performing sensitive operations, such as authentication, installation, or webhook registration, based on guidance alone. The skill emphasizes that live capabilities must be resolved before any API operation, which reduces the risk of accidental obedience to instructions embedded in external documentation.\n- [EXTERNAL_DOWNLOADS]: The skill requires network access to https://cursor.com/docs/api/origin/* to retrieve documentation indices (llms.txt) and reference files (openapi.yaml, llms-full.txt). These downloads are directed to the official product domain and are essential for the skill's primary purpose of providing up-to-date API information.

cursor-origin-api