twitter-cli

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly instructs agents to ask users to copy/paste full browser cookie strings (and then extract and export auth_token/ct0), which requires the LLM to receive and potentially echo or embed secret values verbatim, creating a high exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The codebase is a mostly-legitimate Twitter/X CLI, but it deliberately implements full-browser cookie extraction and includes explicit, instructive guidance (in SKILL.md/README) to ask users or agents to paste full cookie strings into chat — plus anti-detection and proxy guidance — which enables credential harvesting and abuse (high-risk social-engineering / credential-theft vector); there is no hidden exfiltration in the code itself, but the documented agent workflow actively instructs collecting secrets and provides automation to forward/use them.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). This skill (SKILL.md and AGENTS.md) instructs agents to fetch and parse public Twitter/X content (feed, search, tweet detail, articles) and includes agent workflows that read those tweets and then perform actions (e.g., "Like all search results", "Quote-tweet", "Get user info then follow"), meaning untrusted, user-generated third-party content from x.com is ingested and can drive subsequent tool use.