The Agent Skills Directory

[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill requires the installation of the x-developer package from the NPM registry. It also provide instructions to clone the Xquik-dev/x-twitter-scraper repository for inspection. The setup instructions utilize a specific Git commit hash (bfa27fab00dbb8b5367e15153c5723ee608ba00b), which is a security best practice for pinning to a reviewed version of external code.
[COMMAND_EXECUTION]: The documentation includes shell commands for the user or agent to execute, including repository cloning, package installation via NPM, and environment variable configuration for API keys.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The agent retrieves untrusted external data from X (Twitter), such as tweets, user bios, replies, and direct messages, which may contain malicious instructions designed to influence agent behavior.
Boundary markers: The skill contains explicit instructions for the agent to treat the workflow as critical-risk and requires mandatory user approval before performing any "write" actions like posting or following.
Capability inventory: The skill grants the agent the ability to post tweets, send direct messages, follow/unfollow accounts, and register webhooks based on retrieved data.
Sanitization: The skill relies on human-in-the-loop approval as the primary mitigation rather than technical sanitization or filtering of the ingested social media content.

x-twitter-scraper