The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches configuration and contract information from the service's official domain at https://xquik.com/openapi.json to perform runtime validation and ensure API compatibility.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with public X (formerly Twitter) content, which acts as a surface for indirect prompt injection attacks where malicious instructions could be embedded in posts or user profiles.
Ingestion points: Untrusted data is retrieved through the /api/v1/x/tweets/search and timeline endpoints, as implemented in the searchPublicPosts function in examples/core.md.
Boundary markers: The SKILL.md file contains strong instructional warnings for the agent to treat all X content, API errors, and webhook payloads as untrusted external data and to never execute instructions found within them.
Capability inventory: The skill includes capabilities to perform network requests and state-changing actions (creating posts, following users, etc.) on the X platform, though these are mitigated by explicit requirements for user approval.
Sanitization: While the documentation emphasizes treating content as data, the provided TypeScript examples do not demonstrate specific sanitization or escaping of the retrieved strings before they are processed by the agent.

api-search-xquik