EnriWeb MCP Server

An MCP server that provides web search and URL fetching capabilities

README.md

EnriWeb

EnriWeb is a Model Context Protocol (MCP) server over stdio that exposes web search and URL fetching tools by delegating execution to EnriProxy.

If your MCP client can call MCP tools, it can do web search / fetch in a consistent way without implementing provider-specific scraping logic.

What this project is

  • An MCP server process your MCP host launches (OpenCode, Claude Code, Codex, etc.)
  • A thin client for EnriProxy (input validation + structured output)

Requirements

  • Node.js >= 22 (recommended: Node 24 LTS)
  • A reachable EnriProxy server with:
    • POST /v1/tools/web_search
    • POST /v1/tools/web_fetch
  • An EnriProxy API key (configured on the EnriProxy side)

Install

# Global install
npm install -g @bedolla/enriweb

# Or run without installing
npx -y @bedolla/enriweb@latest --help

Build

npm install
npm run typecheck
npm run build

Usage

1) Configure your MCP host

EnriWeb runs as an MCP server over stdio. Your MCP host is responsible for launching the process.

Example: global install

{
  "EnriWeb": {
    "type": "stdio",
    "command": "enriweb",
    "args": [],
    "env": {
      "ENRIPROXY_URL": "http://127.0.0.1:8787",
      "ENRIPROXY_API_KEY": "YOUR_ENRIPROXY_API_KEY"
    }
  }
}

Example: no install (always uses whatever npm currently tags as latest)

{
  "EnriWeb": {
    "type": "stdio",
    "command": "npx",
    "args": ["-y", "@bedolla/enriweb@latest"],
    "env": {
      "ENRIPROXY_URL": "http://127.0.0.1:8787",
      "ENRIPROXY_API_KEY": "YOUR_ENRIPROXY_API_KEY"
    }
  }
}
Use a local dev checkout
{
  "EnriWeb": {
    "type": "stdio",
    "command": "node",
    "args": ["C:\\\\Users\\\\Administrator\\\\Projects\\\\EnriWeb\\\\dist\\\\index.js"],
    "env": {
      "ENRIPROXY_URL": "http://127.0.0.1:8787",
      "ENRIPROXY_API_KEY": "YOUR_ENRIPROXY_API_KEY"
    }
  }
}

Configuration

EnriWeb is configured via environment variables:

  • ENRIPROXY_URL (string, optional, default: http://127.0.0.1:8787)
  • ENRIPROXY_API_KEY (string, required)
  • ENRIWEB_TIMEOUT_MS (string, optional, default: 60000)
    • Parsed as an integer (milliseconds).
  • ENRIWEB_WEB_FETCH_DEFAULT_MAX_CHARS (string, optional, default: 200000)
    • Parsed as an integer.
  • ENRIWEB_GITHUB_TOKEN (string, optional)
    • Used for GitHub API enrichment to improve rate limits.

MCP tools

EnriWeb exposes these MCP tools:

  • web_search
  • web_fetch
Tool inputs (option-by-option)

General notes:

  • All tools accept a single JSON object as their input (the MCP arguments for that tool).
  • EnriWeb returns both:
    • a short human-readable preview (content)
    • the full result payload (structuredContent)

`web_search`

Search the web via EnriProxy.

Inputs:

  • query (string, required): search query string.
  • max_results (number, optional)
    • Must be >= 1.
    • If omitted, EnriProxy uses its configured default.
    • The upper limit is enforced server-side (EnriWeb does not hardcode a max).
  • recency (string, optional, default: noLimit)
    • One of: oneDay | oneWeek | oneMonth | oneYear | noLimit
  • allowed_domains (string[], optional): allowlist of domains to include.
  • blocked_domains (string[], optional): blocklist of domains to exclude.
  • search_prompt (string, optional): extra context to refine the search intent.

Example arguments object:

{
  "query": "qdrant docker compose autostart systemd",
  "max_results": 10,
  "recency": "oneMonth"
}

`web_fetch`

Fetch and read content from a URL via EnriProxy.

Inputs:

  • url (string, required unless cursor is provided): full URL (http:// or https://).
  • cursor (string, optional): opaque cursor returned by a previous web_fetch call.
  • offset (number, optional, default: 0): cursor read offset in characters.
  • limit (number, optional): cursor read limit in characters (default: max_chars).
  • prompt (string, optional): extraction hint (what to focus on).
  • max_chars (number, optional): maximum content length (default: ENRIWEB_WEB_FETCH_DEFAULT_MAX_CHARS).

Notes:

  • If the response includes a cursor, you can page through the captured content by calling web_fetch again with cursor + offset + limit.

Example arguments object:

{
  "url": "https://example.com/docs",
  "max_chars": 200000
}

Tools 2

web_searchSearch the web via EnriProxy.
web_fetchFetch and read content from a URL via EnriProxy.

Environment Variables

ENRIPROXY_URLThe URL of the EnriProxy server
ENRIPROXY_API_KEYrequiredAPI key for EnriProxy authentication
ENRIWEB_TIMEOUT_MSTimeout in milliseconds for requests
ENRIWEB_WEB_FETCH_DEFAULT_MAX_CHARSDefault maximum content length for web fetches
ENRIWEB_GITHUB_TOKENGitHub API token for rate limit improvement

Try it

Search for the latest documentation on Qdrant docker compose setup from the last month.
Fetch the content of https://example.com/docs and summarize the main features.
Search for recent news about AI regulations, limiting results to tech news domains.
Fetch the content of the provided URL and focus on extracting the installation instructions.

Frequently Asked Questions

What are the key features of EnriWeb?

Provides web search capabilities via EnriProxy. Supports URL content fetching with extraction hints. Allows filtering by domain and recency. Supports pagination for large content via cursors. Provides both human-readable previews and structured content.

What can I use EnriWeb for?

Automating research tasks by searching for specific technical documentation.. Extracting and summarizing content from web pages for AI-assisted coding.. Monitoring specific domains for updates using filtered search queries.. Retrieving large web documents in chunks for analysis by an AI agent..

How do I install EnriWeb?

Install EnriWeb by running: npm install -g @bedolla/enriweb

What MCP clients work with EnriWeb?

EnriWeb works with any MCP-compatible client including Claude Desktop, Claude Code, Cursor, and other editors with MCP support.

Turn this server into reusable context

Keep EnriWeb docs, env vars, and workflow notes in Conare so your agent carries them across sessions.

Open Conare