← Blog

119 Free MCP Tools for AI Agents — No API Keys Required

Pipeworx hosts 119 MCP tool packs covering weather, science, finance, gaming, and more. Connect any AI agent in one line — free, no API keys.

AI agents are only as useful as the data they can reach. A language model that can reason about anything but only knows what was in its training set is a calculator without batteries. The hard part has never been the reasoning — it’s been wiring up the data.

Pipeworx solves that. We host 119 MCP tool packs across every category imaginable: live weather, scientific literature, government databases, gaming universes, developer tools, and more. Every pack is available through a single gateway URL, with no API keys to manage, no SDKs to install, and no authentication to configure. You pick the tools you need and start querying.

What is MCP?

The Model Context Protocol (MCP) is an open standard, introduced by Anthropic in late 2024, that gives AI agents a structured way to call external tools. Instead of every AI app inventing its own function-calling format, MCP defines a universal JSON-RPC interface: the agent sends a tools/call request, the server executes it and returns structured data, and the agent folds that data into its reasoning. Claude, GPT-4o, Gemini, and most open-source models now support MCP natively. It is, essentially, USB for AI tools.

Quick Start

Claude Desktop

Add Pipeworx to your claude_desktop_config.json (found at ~/Library/Application Support/Claude/ on macOS):

{
  "mcpServers": {
    "pipeworx": {
      "command": "npx",
      "args": ["-y", "@pipeworx/gateway"],
      "env": {}
    }
  }
}

Restart Claude Desktop and all 119 packs appear immediately as available tools.

CLI

npx pipeworx use weather

This fetches the weather pack manifest and prints a ready-to-paste MCP config block for any client.

Direct gateway

https://gateway.pipeworx.io/mcp

Point any MCP-compatible client at this URL. No authentication required for the free tier (50 calls/day per IP).

Pack Highlights

Here are roughly 20 of the strongest packs, grouped by what they’re actually useful for.

Science and Research

crossref — Query 160 million scholarly works by DOI, title, or author. Returns metadata, citation counts, and abstract text. Essential for any research assistant pipeline.

openalex — Open Access layer on top of the academic graph. Search institutions, authors, concepts, and funding sources. Useful for mapping research landscapes.

pubmed — Search the National Library of Medicine’s 35 million biomedical abstracts. Returns structured records with MeSH terms, journal info, and links to full text.

gbif — Global Biodiversity Information Facility. Query species occurrence records, taxonomy trees, and specimen data. Covers 2 billion occurrence records from museums and field surveys worldwide.

Government and Public Data

treasury — U.S. Treasury data including exchange rates, federal debt, and spending records. The pack wraps the fiscal data API with clean typed responses.

fbiwanted — Live query against the FBI’s Most Wanted list. Returns name, charges, physical description, and reward information. Useful for identity verification workflows.

fda — Food and Drug Administration drug database. Query approved drugs, recalls, adverse events, and clinical study references.

ukpolice — UK Police API: crime statistics by postcode, stop-and-search records, outcomes, and neighborhood team contacts. Covers England, Wales, and Northern Ireland.

Entertainment and Gaming

pokemon — Full PokeAPI wrapper. Fetch stats, moves, evolutions, abilities, and encounter locations for all 1,000+ Pokemon across every generation. Highly popular with younger users testing agents for the first time.

rickmorty — Characters, episodes, and locations from Rick and Morty. Returns canonical show data useful for building fandom apps or testing agent memory.

dnd5e — Dungeons and Dragons 5th Edition rules data: spells, classes, equipment, monsters, and ability checks. The pack is used by several tabletop assistant apps in production.

swapi — Star Wars API. Planets, starships, films, species, and characters from the main saga. Well-structured reference data that works well as a demonstration dataset.

chess — Lichess integration. Fetch game records by username, query opening theory, look up player ratings. Real-time data updated continuously.

Developer Tools

github — Repository metadata, file contents, issues, pull requests, commit history, and contributor stats. Does not require a personal access token for public repos.

npm — Package registry search, version history, dependency trees, and download statistics. Useful for dependency auditing agents.

stackexchange — Search across Stack Overflow, Server Fault, Super User, and 170+ other sites. Returns accepted answers with vote counts and author metadata.

nvd — National Vulnerability Database. Query CVEs by ID, CVSS score range, affected product, or publication date. Indispensable for security-aware agents.

Real-Time Data

weather — Current conditions and forecasts via Open-Meteo. Returns temperature, precipitation probability, wind, UV index, and hourly breakdowns for any coordinate. No key required.

flights — Flight status, route data, and airport information. Useful for travel assistant applications.

sports — Live scores, standings, and schedules for major leagues including NFL, NBA, MLB, NHL, and Premier League.

spacex — SpaceX launch history and upcoming missions. Detailed records including rocket, payload, launch site, and outcome. The dataset is frozen but comprehensive.

launches — The Launch Library 2 API, covering all orbital and suborbital launch providers worldwide. More comprehensive than the SpaceX pack for multi-agency coverage.

Reference and Language

wikipedia — Article summaries, introductory paragraphs, and disambiguation results via the Wikipedia REST API. Returns clean plain text, not raw wikitext.

words — Word definitions, synonyms, antonyms, pronunciation, frequency scores, and usage examples from the Wordnik dataset.

translate — Language detection and translation across 100+ languages via LibreTranslate. Runs on open infrastructure.

dictionary — The Free Dictionary API. Phonetics, part of speech, definitions, and example sentences. Complements words with different coverage.

Fun

jokes — Programming, dark, pun, and general jokes from the JokeAPI. Supports category filtering and safe-mode for appropriate content.

dadjokes — Exclusively dad jokes, sourced from icanhazdadjoke.com. Single-purpose but surprisingly requested.

chucknorris — Chuck Norris facts API. Categorized, searchable, and genuinely useful for testing that your agent can handle informal text.

catfacts — Random facts about cats. The pack most frequently used in “hello world” MCP tutorials.

xkcd — Fetch any xkcd comic by number, or get the latest. Returns title, image URL, alt text, and transcript.

Try It Live

Before writing any configuration, you can test any pack in the browser at pipeworx.io/packs/weather. The playground lets you call tools directly, inspect raw JSON responses, and copy the exact invocation for use in your own agent.

How to Connect

MCP JSON config (any client)

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp",
      "transport": "http"
    }
  }
}

Claude Code (CLI)

claude mcp add pipeworx --url https://gateway.pipeworx.io/mcp

curl — direct tool call

curl -X POST https://gateway.pipeworx.io/mcp \
  -H 'Content-Type: application/json' \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "weather_get_forecast",
      "arguments": {
        "latitude": 37.7749,
        "longitude": -122.4194,
        "days": 3
      }
    }
  }'

Example response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "content": [
      {
        "type": "text",
        "text": "San Francisco forecast for the next 3 days:\n\nMon Apr 7: High 16°C / Low 11°C. Partly cloudy. Precipitation 12%.\nTue Apr 8: High 14°C / Low 10°C. Fog in the morning, clearing by noon. Precipitation 8%.\nWed Apr 9: High 17°C / Low 12°C. Mostly sunny. Precipitation 3%."
      }
    ]
  }
}

The gateway handles all upstream authentication, rate limiting, and error normalization. Your agent gets clean, structured text back every time.

Open Source

Every Pipeworx tool pack is open source under the MIT license. The code lives at github.com/pipeworx-io — one repo per pack, each independently publishable to npm as @pipeworx/mcp-<slug>.

If a pack has a bug, you can fix it and submit a pull request. If the upstream API changes its schema, you can patch it. If you want to fork and self-host, everything you need is there.

We also publish each pack to the official MCP Registry at registry.modelcontextprotocol.io so it shows up in ecosystem tooling regardless of whether you use the Pipeworx gateway.

What’s Next

The 119 packs currently live are just the start. We’re actively building out:

  • More packs — Community submissions are open. If there’s a public API you want wrapped as an MCP pack, submit it here and we’ll review and publish it.
  • Authenticated packs — Some APIs require keys (financial data, premium weather, licensed content). We’re building a proxy layer that lets you supply your own keys through the gateway without exposing them client-side.
  • Pack versioning — Semantic versions for every pack so you can pin agents to a stable schema.
  • Usage analytics — A dashboard showing which tools your agents call most, latency distributions, and error rates.
  • Team accounts — Higher rate limits, shared API keys, and usage attribution across agents.

The goal is simple: no AI agent should have to go without access to the data it needs because the plumbing was too hard to set up. Every public API should be one line of config away.

If you build something with Pipeworx, we’d love to hear about it. Drop us a note through the feedback widget or open an issue on GitHub.