Agent Toolbox

<div align="center"> # ⚡ Agent Toolbox **13 production-ready tools for AI agents — one API key, zero complexity.** [](https://www.npmjs.com/package/agent-toolbox-mcp) [](https://pypi.org/project/langchain-agent-toolbox/) [](LICENSE) [](https://api.toolboxlite.com/docs) [](https://glama.ai/mcp/servers/@Vincentwei1021/agent-toolbox) [Docs](https://api.toolboxlite.com/docs) · [Playground](https://api.toolboxlite.com/playground) · [OpenAPI Spec](https://api.toolboxlite.com/openapi.json) · [npm](https://www.npmjs.com/package/agent-toolbox-mcp) </div> --- ## Why Agent Toolbox? Building AI agents that interact with the real world? You'd normally need: | Without Agent Toolbox | With Agent Toolbox | |---|---| | 6+ API keys (Google, OpenWeatherMap, Yahoo Finance, ...) | **1 API key** | | Different auth methods per service | **Unified Bearer auth** | | Parse different response formats | **Consistent JSON responses** | | Handle rate limits per provider | **Built-in rate limiting** | | No caching, redundant calls | **Smart LRU caching** | | Pay $50-200+/mo for APIs | **Free tier: 1,000 calls/mo** | ## Quick Start (30 seconds) ### 1. Get your free API key ```bash curl -X POST https://api.toolboxlite.com/v1/auth/register \ -H "Content-Type: application/json" \ -d '{"email": "you@example.com"}' ``` ### 2. Make your first call ```bash curl -X POST https://api.toolboxlite.com/v1/search \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{"query": "best AI agent frameworks"}' ``` ### 3. That's it! 🎉 No credit card. No setup. 13 endpoints ready to use. ## 13 Endpoints | Endpoint | Description | Cache | |----------|-------------|-------| | `POST /v1/search` | Web search via DuckDuckGo | 5 min | | `POST /v1/extract` | Extract content from any URL | 5 min | | `POST /v1/screenshot` | Capture webpage screenshots | 1 hr | | `POST /v1/weather` | Weather & forecast | 15 min | | `POST /v1/finance` | Stock quotes & exchange rates | 15 min | | `POST /v1/validate-email` | Email validation (MX + SMTP) | 1 hr | | `POST /v1/translate` | Translate 100+ languages | 1 hr | | `POST /v1/geoip` | IP geolocation lookup | 1 hr | | `POST /v1/news` | News article search | 5 min | | `POST /v1/whois` | Domain WHOIS lookup | 1 hr | | `POST /v1/dns` | DNS record queries | 5 min | | `POST /v1/pdf-extract` | Extract text from PDFs | 1 hr | | `POST /v1/qr` | Generate QR codes | 24 hr | ## MCP Server Use Agent Toolbox as an MCP server in Claude Desktop, Cursor, or Windsurf: ```bash npm install -g agent-toolbox-mcp ``` Add to Claude Desktop config: ```json { "mcpServers": { "agent-toolbox": { "command": "agent-toolbox-mcp" } } } ``` See [MCP Quickstart Guide](examples/mcp-quickstart/) for full setup instructions. ## SDK Integrations ### Python — LangChain ```bash pip install langchain-agent-toolbox ``` ```python from langchain_agent_toolbox import AgentToolboxSearchTool search = AgentToolboxSearchTool() result = search.invoke({"query": "AI agents", "count": 3}) ``` ### Python — LlamaIndex ```bash pip install llamaindex-agent-toolbox ``` ```python from llamaindex_agent_toolbox import AgentToolboxSearchTool search = AgentToolboxSearchTool() result = search.call(query="AI agents", count=3) ``` ## Examples | Example | Description | |---------|-------------| | [Research Agent](examples/research-agent/) | Auto-research any topic → markdown report | | [Website Monitor](examples/website-monitor/) | Monitor sites with DNS + screenshots + content | | [MCP Quickstart](examples/mcp-quickstart/) | Claude Desktop / Cursor / Windsurf setup | ## Pricing | Plan | Price | Calls/month | Rate Limit | |------|-------|-------------|------------| | **Free** | $0 | 1,000 | 60/min | | **Pro** | $29/mo | 50,000 | 120/min | | **Scale** | $99/mo | 500,000 | 300/min | All plans include all 13 endpoints. No feature gating. ## Self-Hosting ```bash git clone https://github.com/Vincentwei1021/agent-toolbox.git cd agent-toolbox npm install && npm run build cp .env.example .env # Add your config npm start ``` Or with Docker: ```bash docker build -t agent-toolbox . docker run -p 3100:3100 --env-file .env agent-toolbox ``` ## Used By > *Building with Agent Toolbox? [Open an issue](https://github.com/Vincentwei1021/agent-toolbox/issues) to get listed here.* ## License MIT © [Vincentwei1021](https://github.com/Vincentwei1021)