MCP Internet Server

Small MCP server exposing two tools for internet access:

• search_internet(query, max_results=5) — perform a web search and return structured results.

• fetch_webpage_content(url) — fetch and extract text from an HTML page with SSRF protections.

Features

• Structured (dict) responses for easier downstream processing.

• SSRF protections: optional ALLOWED_DOMAINS allowlist and private IP blocking.

• JSON-line audit logging (path configurable via AUDIT_LOG_PATH).

• Unit tests using pytest.

Quick start

1. Create and activate a virtualenv (recommended):

python -m venv .venv
# Windows
.venv\Scripts\activate
# Unix/macOS
source .venv/bin/activate

2. Install dependencies:

python -m pip install -r requirements.txt

3. Run a quick smoke test:

python run_test.py

4. Run unit tests:

python -m pytest -q

5. Run test coverage:

python -m pytest --cov=server --cov-report=term-missing

6. MCP json to use in inference engine

{
  "mcpServers": {
    "internet_assistant": {
      "command": "PATH TO YOUR PYTHON EXECUTABLE",
      "args": [
        "PATH TO YOUR server.py"
      ]
    }
  }
}

Configuration

• ALLOWED_DOMAINS: optional comma-separated list of domains that fetch_webpage_content is allowed to access. If unset, fetching is allowed for any public hostname (subject to IP checks).

• AUDIT_LOG_PATH: path to write newline-delimited JSON audit events (default: audit.log).

• METRICS_PORT: port to expose Prometheus metrics via HTTP. Set to 0 or unset to disable metrics export.

Notes

• The project prefers the ddgs package for search; duckduckgo_search fallback is supported.

• Audit log rotation, remote logging, or more advanced rate-limiting are not included but recommended for production deployments.

Files of interest

• server.py — main MCP tools implementation

• run_test.py — smoke tests

• tests/test_server.py — pytest unit tests

• requirements.txt — dependencies

License: MIT