wet — Claude Code MCP server: what it does, how to install
Gives agents web search through SearXNG, multi-step page extraction, and version-aware library documentation search.
claude mcp add wet -e API_KEYS=<API_KEYS> -e GITHUB_TOKEN=<GITHUB_TOKEN> -- uvx wet-mcpWhat wet does
The server consolidates several retrieval mechanisms for coding agents. It runs an embedded SearXNG engine to aggregate web results with response caching and snippet limits. For academic research, it queries sources such as arXiv, PubMed, and CrossRef. Documentation retrieval relies on SQLite FTS5 hybrid search combined with HyDE.
Content extraction includes browser interactions. Agents can click elements and submit forms through patchright, maintaining session state when needed.
The PyPI distribution is named wet-mcp. If you need image or audio processing, this tool will not help: the developers removed media.analyze in version 2.0.0.
What it helps with
- run web queries across search engines using SearXNG
- pull academic research papers from arXiv and PubMed
- search technical documentation pinned to specific library versions
- fill out web forms and click page elements with patchright
Example prompts for the agent
- Search arXiv for recent publications on sparse autoencoders.
- Look up FastAPI documentation for route dependency overrides.
- Navigate to the portal, enter the credentials, and submit the login form.
How to install wet
Platform CLI
claude mcp add wet -e API_KEYS=<API_KEYS> -e GITHUB_TOKEN=<GITHUB_TOKEN> -- uvx wet-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from wet-mcp <command>
Platform CLI
codex mcp add wet --env API_KEYS=<API_KEYS> --env GITHUB_TOKEN=<GITHUB_TOKEN> -- uvx wet-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from wet-mcp <command>
Config file
{
"mcpServers": {
"wet": {
"command": "uvx",
"args": [
"wet-mcp"
],
"env": {
"API_KEYS": "<API_KEYS>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from wet-mcp <command>
Config file
{
"servers": {
"wet": {
"type": "stdio",
"command": "uvx",
"args": [
"wet-mcp"
],
"env": {
"API_KEYS": "<API_KEYS>",
"GITHUB_TOKEN": "<GITHUB_TOKEN>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from wet-mcp <command>
Platform CLI
gemini mcp add -e API_KEYS=<API_KEYS> -e GITHUB_TOKEN=<GITHUB_TOKEN> wet uvx wet-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from wet-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about wet
How to install wet in Claude Code?
Run claude mcp add wet -e API_KEYS=<API_KEYS> -e GITHUB_TOKEN=<GITHUB_TOKEN> -- uvx wet-mcp in your terminal.
What does the wet MCP server do?
It supplies web search via SearXNG, searches academic papers, and fetches documentation chunks for libraries.
Can wet process images or video?
No. The media.analyze feature was removed in release 2.0.0, so the server only handles text search, web navigation, and documentation.
