io.github.cyanheads/guardian-mcp-server

@cyanheads/guardian-mcp-server

Search, browse, and retrieve full article text from The Guardian's journalism archive (1999–present) via MCP. STDIO or Streamable HTTP.

3 Tools

Overview

The Guardian's journalism archive, 1999 to today. Search full text, browse by section or tag, and fetch individual articles.

Tools

Capability reference

guardian_search_articles tool

• Boolean query syntax (AND/OR/NOT, quoted phrases), optional section/tag/contributor and from_date/to_date (YYYY-MM-DD) filters; order_by: relevance (default), newest, or oldest; page + page_size (1–50, default 10)
• Articles carry metadata, HTML-stripped body text and a truncated flag; results include total, page, pages, page_size, and order_by
• Fails with no_results or invalid_date (dates must be real calendar dates)
• tag accepts comma (AND), pipe | (OR), hyphen - (NOT), and parentheses, e.g. environment/climate-crisis|environment/flooding. Wrap an OR expression in parentheses when also passing contributor.

guardian_get_article tool

• Required article_id: the path-slug id returned by guardian_search_articles or guardian_browse_content
• Returns HTML-stripped body text, metadata, contributors, pillar/section classification, and truncated
• Fails with not_found

guardian_browse_content tool

• Four mode values: section_latest (requires section_id), tag_latest (requires tag_id), list_sections, list_tags; page + page_size (1–50, default 10) for articles and tags
• list_tags filters by optional query and tag_type (keyword, contributor, blog, series, tone, type, publication, newspaper-book, newspaper-book-section); output carries mode, total, and the matching results/sections/tags array
• Fails with missing_section_id, missing_tag_id, section_not_found, or tag_not_found. tag_id accepts the same boolean expressions as search's tag.

All three tools also fail with status-specific HTTP errors. Not retryable: unauthorized (401), access_denied (403), request_rejected (400), validation_failed (422), invalid_request, request_conflict, request_not_found, and api_unsupported (501). Retryable: api_timeout (408/425/504), rate_limited (429), and api_error (other 5xx). Each error carries a recovery hint.

Features

Built on @cyanheads/mcp-ts-core: stdio and Streamable HTTP transports, pluggable auth (none / jwt / oauth), swappable storage (in-memory, filesystem, Supabase, Cloudflare KV/R2/D1), structured logging with optional OpenTelemetry tracing.

Guardian-specific:

• Wraps the Guardian Open Platform API with a free non-commercial developer key
• Full body text extraction — HTML stripped, not just headlines or abstracts
• Contributor ID discovery via guardian_browse_content mode list_tags with tag_type=contributor
• Section and tag taxonomy browsing (list_sections, list_tags) for filter discovery before searching
• Free tier: 5,000 requests/day, 12 calls/second — the server applies no additional throttling

Agent-friendly output:

• Search and article-returning browse modes cap previews at 2,000 words and set truncated; call guardian_get_article for the full text
• Typed error reasons across all three tools, each paired with a recovery hint
• total / page / pages on every paginated response so callers can track result scope

Getting started

Add the following to your MCP client configuration file. Register for a free Guardian Open Platform API key at open-platform.theguardian.com/access.

{
  "mcpServers": {
    "guardian-mcp-server": {
      "type": "stdio",
      "command": "bunx",
      "args": ["@cyanheads/guardian-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info",
        "GUARDIAN_API_KEY": "your-api-key"
      }
    }
  }
}

Or with npx (no Bun required):

{
  "mcpServers": {
    "guardian-mcp-server": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@cyanheads/guardian-mcp-server@latest"],
      "env": {
        "MCP_TRANSPORT_TYPE": "stdio",
        "MCP_LOG_LEVEL": "info",
        "GUARDIAN_API_KEY": "your-api-key"
      }
    }
  }
}

Or with Docker:

{
  "mcpServers": {
    "guardian-mcp-server": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run", "-i", "--rm",
        "-e", "MCP_TRANSPORT_TYPE=stdio",
        "-e", "GUARDIAN_API_KEY=your-api-key",
        "ghcr.io/cyanheads/guardian-mcp-server:latest"
      ]
    }
  }
}

For Streamable HTTP, set the transport and start the server:

MCP_TRANSPORT_TYPE=http MCP_HTTP_PORT=3010 GUARDIAN_API_KEY=your-api-key bun run start:http
# Server listens at http://localhost:3010/mcp

Prerequisites

• Bun v1.4.0 or higher (or Node.js v24+).
• A free Guardian Open Platform API key — register at open-platform.theguardian.com/access. The non-commercial developer tier is free and instant.

Installation

1. Clone the repository:

git clone https://github.com/cyanheads/guardian-mcp-server.git

1. Navigate into the directory:

cd guardian-mcp-server

1. Install dependencies:

bun install

1. Configure environment:

cp .env.example .env
# Edit .env and set GUARDIAN_API_KEY

Configuration

See .env.example for the full list of optional overrides.

Running the server

Local development

• Build and run:

  # One-time build
  bun run rebuild
  
  # Run the built server
  bun run start:stdio
  # or
  bun run start:http

• Run checks and tests:

  bun run devcheck   # Lint, format, typecheck, security
  bun run test       # Vitest test suite
  bun run lint:mcp   # Validate MCP definitions against spec

Docker

docker build -t guardian-mcp-server .
docker run --rm -e GUARDIAN_API_KEY=your-api-key -p 3010:3010 guardian-mcp-server

The Dockerfile defaults to HTTP transport, stateless session mode, and logs to /var/log/guardian-mcp-server. OpenTelemetry peer dependencies are installed by default — build with --build-arg OTEL_ENABLED=false to omit them.

Project structure

Development guide

See CLAUDE.md / AGENTS.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for request-scoped logging, ctx.state for tenant-scoped storage
• Register new tools in the createApp() tools array in src/index.ts
• Wrap external API calls: validate raw → normalize to domain type → return output schema; never fabricate missing fields

Contributing

Issues are welcome. Run checks and tests before submitting:

bun run devcheck
bun run test

License

Apache-2.0 — see LICENSE for details.

Powered by The Guardian.