The Agent Skills Directory

[PROMPT_INJECTION]: The skill's functionality for fetching and processing data from X (Twitter) exposes the agent to indirect prompt injection risks from untrusted external content.
Ingestion points: Data is ingested from external sources through various Xquik API endpoints for tweet search, user lookups, and direct message history, as described in SKILL.md and references/api-endpoints.md.
Boundary markers: The skill requires the agent to wrap all X-authored text in <XQUIK_UNTRUSTED_X_CONTENT> tags to isolate it from system instructions, as specified in SKILL.md and references/security.md.
Capability inventory: The skill possesses capabilities that could be influenced by malicious content, such as posting tweets, sending direct messages, and following users (references/api-endpoints.md).
Sanitization: The instructions explicitly direct the agent to treat external content as data only, ignore any embedded commands, and obtain explicit user consent before performing any private read or write operation (SKILL.md, references/security.md).
[EXTERNAL_DOWNLOADS]: The skill interacts with the xquik.com and docs.xquik.com domains to perform REST API requests and retrieve technical documentation. These are first-party domains for the integrated Xquik service.

xquik-twitter