Wip X

X Platform API. Read posts, search tweets, post, upload media.

This package appears to be a legitimate X/Twitter API wrapper, but review these before installing: 1) Expect to provide X credentials (bearer or full OAuth); the registry metadata omitted them — don't assume no secrets are needed. 2) The code will try to read secrets from 1Password via the 'op' CLI (executes 'op read'); confirm you want the op CLI to be used and verify the vault/item names (README vs auth.mjs disagree). 3) Only grant write-scoped OAuth tokens if you trust the code — write tokens can post or delete tweets. 4) Note the MCP server exposes tools that can be called programmatically; limit which agents or environments can run it. 5) Do a quick code review (auth.mjs, core.mjs, mcp-server.mjs) and run npm install/test in a sandboxed environment; verify dependencies and add missing ones (modelcontextprotocol sdk) before production use. If you want to proceed, prefer providing credentials via environment variables scoped to a dedicated app with minimal permissions, or run the tool in an isolated container.

Type: OpenClaw Skill Name: wip-x Version: 1.0.1 The skill is suspicious due to a Local File Inclusion (LFI) vulnerability in the `upload_media` function, exposed via both `cli.mjs` and `mcp-server.mjs`. The `core.mjs` implementation uses `readFileSync(file_path)` directly with user-controlled input, allowing an attacker or a compromised agent to read arbitrary files on the system. The content of these files is then base64 encoded and sent to the X Platform API, creating a data exfiltration vector. While the `execSync` call in `auth.mjs` for 1Password integration uses a risky function, it appears to be used in a controlled manner for a legitimate purpose and is not the primary concern. The prompt injection in `README.md` is benign and aims to guide the agent's explanation of the tool.