Agent Web Search: MCP tool for Hermes Agent

Provider-neutral web search for AI agents as an MCP server, CLI and Python library

What Agent Web Search does

Agent Web Search, published on PyPI as agent-web-search-mcp, routes natural-language questions to search backends built for agents instead of fanning keyword queries out to Google, Bing or Baidu. It combines model-native grounding (Responses, ARK) with agent-oriented search APIs such as Exa and Parallel, and uses DDGS as its only conventional search backend. Results come back as answers, citations or structured evidence that an agent can use directly.

An agent can reach the same search core in three ways: a native MCP tool over stdio or Streamable HTTP, a native plugin for Hermes or DeepSeek Harness, or the agent-web-search CLI taught through a standard Agent Skill. The default providers (DDGS, Exa, Parallel) need no API key. The README lists Codex CLI, Claude Code, OpenCode, Hermes and DeepSeek Harness as supported clients, and a Simplified Chinese README is included.

Key features

MCP server with stdio and Streamable HTTP transports
Native plugins for Hermes and DeepSeek Harness
agent-web-search CLI with an accompanying Agent Skill
Keyless default providers: DDGS, Exa and Parallel
Model-native grounding through Responses and ARK
Python library interface

When to use it

Giving a Hermes Agent live web search without registering for a search API key
Sharing one search setup across Hermes, Claude Code and Codex CLI
Getting cited answers from model-grounded search inside an agent task
Serving search to remote MCP clients over Streamable HTTP

Who it is for: Developers who run Hermes Agent or other coding agents and want a single web search tool that works across them.

How it fits with Hermes Agent

Hermes is one of several supported agents, with a native Hermes plugin alongside the MCP and CLI routes.

How to install Agent Web Search

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

pipx install agent-web-search-mcp
python -m pip install agent-web-search-mcp

Requirements: Python 3.10+; the default providers (DDGS, Exa, Parallel) need no API key

FAQ

What is Agent Web Search?

Agent Web Search is an open-source MCP server, CLI and Python library that provides web search for AI agents. It aggregates model-native grounding and agent search APIs behind one provider-neutral contract.

Does Agent Web Search work with Hermes Agent?

Yes. Hermes is listed as a supported client, with a native Hermes plugin as one of the integration options alongside MCP and the CLI.

How do I install Agent Web Search?

Install it with pipx install agent-web-search-mcp, or use python -m pip install agent-web-search-mcp inside an active Python environment. The README also shows trying it without installing through uvx.