The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is designed to access and extract sensitive authentication tokens directly from local browser storage (Arc, Chrome, Edge, Firefox, Brave). It also explicitly instructs the user to unlock the macOS keychain ('~/Library/Keychains/login.keychain-db') to allow the tool to decrypt stored browser credentials.- [EXTERNAL_DOWNLOADS]: The installation process involves downloading the 'twitter-cli' package from PyPI or cloning it from a public GitHub repository ('github.com/jackwener/twitter-cli').- [COMMAND_EXECUTION]: The skill provides Python implementation examples that use 'subprocess.run' to programmatically execute CLI commands and handle Twitter data, which can lead to command injection if tweet content is not properly sanitized before being passed to shell commands.- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from Twitter (feeds, search results, and bookmarks) into the agent context. Because the skill also possesses 'write' capabilities (posting, replying, following), it is vulnerable to indirect prompt injection where malicious instructions embedded in a tweet could trigger unauthorized actions. Ingestion points: 'twitter feed', 'twitter search'; Boundary markers: Absent; Capability inventory: 'twitter post', 'twitter follow', 'twitter delete'; Sanitization: Absent.

twitter-cli-skill