From 22e9f1a5a2c586bd303c5bc3d6da93b7f9d4ff0a Mon Sep 17 00:00:00 2001 From: Usman Aslam <53818205+usmanaslam712@users.noreply.github.com> Date: Fri, 11 Jul 2025 03:13:41 +0500 Subject: [PATCH 1/2] Create README.md for Fastn UCL --- fastn-ucl/README.md | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+) create mode 100644 fastn-ucl/README.md diff --git a/fastn-ucl/README.md b/fastn-ucl/README.md new file mode 100644 index 0000000000..aa7628c49d --- /dev/null +++ b/fastn-ucl/README.md @@ -0,0 +1,38 @@ +# Fastn UCL – Model Context Protocol Server + +[Fastn UCL](https://ucl.dev) is a production-ready implementation of the [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol/spec) designed for AI builders. + +It exposes a unified `/command` endpoint that allows developers to interact with 1000+ applications (e.g. Notion, Slack, Gmail, GitHub) via consistent command + context payloads — without repetitive API setup. + +--- + +## 🚀 What is Fastn UCL? + +Fastn UCL (Unified Command Layer) is a multi-tenant MCP that enables: + +- Reusable commands across different tools +- Context-aware execution (ideal for AI agents) +- Embedding directly into agents, SaaS products, and more. +- Built for multi-tenant, multi-agent orchestration + +It follows the MCP spec out of the box. + +📘 [Read the full documentation](https://docs.fastn.ai/ucl-unified-command-layer/about-fastn-ucl) + +--- + +## 🔌 Endpoint + +**`POST /command`** + +Send commands and contexts using a standard MCP payload. + +### Example Request +```json +{ + "command": "send_slack_message", + "context": { + "channel": "#general", + "message": "Hello from Fastn UCL!" + } +} From bb08dcae8e3ba69ad56c395a67883fe1a0680bbb Mon Sep 17 00:00:00 2001 From: Usman Aslam <53818205+usmanaslam712@users.noreply.github.com> Date: Tue, 15 Jul 2025 02:00:25 +0500 Subject: [PATCH 2/2] Update README.md new updated readme --- fastn-ucl/README.md | 141 +++++++++++++++++++++++++++++++++++++------- 1 file changed, 119 insertions(+), 22 deletions(-) diff --git a/fastn-ucl/README.md b/fastn-ucl/README.md index aa7628c49d..16d89791d0 100644 --- a/fastn-ucl/README.md +++ b/fastn-ucl/README.md @@ -1,38 +1,135 @@ # Fastn UCL – Model Context Protocol Server -[Fastn UCL](https://ucl.dev) is a production-ready implementation of the [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol/spec) designed for AI builders. +Fastn UCL is a production-ready, multi-tenant implementation of the **Model Context Protocol (MCP)**, built to help AI agents take real actions across 1000+ tools via a single, unified interface. -It exposes a unified `/command` endpoint that allows developers to interact with 1000+ applications (e.g. Notion, Slack, Gmail, GitHub) via consistent command + context payloads — without repetitive API setup. +With just one /command endpoint, your AI agents can access external services like **Notion, Slack, Gmail, Jira,** and many more, without writing custom glue code, managing auth, or duplicating integration logic. ---- +## 🔧 What is Fastn UCL? -## 🚀 What is Fastn UCL? +**Fastn UCL (Unified Command Layer)** is a secure, MCP-compliant platform that sits between your AI agents and third-party tools. It enables: -Fastn UCL (Unified Command Layer) is a multi-tenant MCP that enables: +- **Reusable Commands:** One command schema works across all tools +- **Context-Aware Execution:** Every action is scoped to the correct tenant, user, and configuration +- **Embedded Integration:** Use directly inside agents or SaaS products, no redirect flows +- **Enterprise-Grade Multitenancy:** Fully isolated tenants within a single deployment +- **Built-in Observability:** Automatic logging, retries, and audit trails -- Reusable commands across different tools -- Context-aware execution (ideal for AI agents) -- Embedding directly into agents, SaaS products, and more. -- Built for multi-tenant, multi-agent orchestration +Whether you're building an AI-native SaaS product or integrating agents into internal tools, Fastn UCL handles the complexity of real-world execution at scale. -It follows the MCP spec out of the box. +## 🌐 Unified Endpoint -📘 [Read the full documentation](https://docs.fastn.ai/ucl-unified-command-layer/about-fastn-ucl) +Fastn UCL exposes a single standardized endpoint: ---- +``` +POST /command +``` -## 🔌 Endpoint +This endpoint receives a payload containing: +- A command: the structured action you want to execute +- A context: all necessary input like credentials, parameters, or metadata -**`POST /command`** +## 📦 Example Request -Send commands and contexts using a standard MCP payload. - -### Example Request ```json { - "command": "send_slack_message", - "context": { - "channel": "#general", - "message": "Hello from Fastn UCL!" - } +"command": "send_slack_message", +"context": { + "channel": "#general", + "message": "Hello from Fastn UCL!" +} } +``` + +Fastn UCL will: +- Identify the correct **tenant and user** +- Match the command to the right **connector configuration** +- Authenticate and route the request to the external tool +- Return a standardized, structured result + +## 🤝 Seamless AI Integration + +Use Fastn UCL with any AI framework: + +| Platform | Use Case Example | +|----------|------------------| +| CrewAI | Multi-agent workflows accessing real-world tools | +| LangChain | LLM apps with tool-use via MCP adapter | +| LlamaIndex | RAG apps that both retrieve info and trigger actions | +| LiveKit | Voice-driven agents interacting with external systems | + +Each sample follows a common structure: + +**User Input → AI Platform → Fastn UCL → External Tool → Output** + +## 🧰 Highlights & Benefits + +- **1000+ Tools Supported:** From Slack and Notion to Salesforce and Active Directory +- **Multitenant by Design:** One deployment, many securely isolated customers +- **No Hosting Needed:** Fully managed platform with zero infra burden +- **Observability Built In:** Real-time monitoring, structured logs, error handling +- **Secure & Compliant:** SOC 2 Type II, GDPR, HIPAA-ready + +## 🚀 Quick Start + +**Prerequisites** +- Python 3.8+ +- Fastn UCL API Key and Space ID (sign up at [ucl.dev](https://ucl.dev/)) +- Optional: OpenAI key (for AI-powered use cases) + +**Steps** +```bash +# Clone any integration sample +cd crewai/ + +# Setup virtual environment +python -m venv venv && source venv/bin/activate + +# Install dependencies +pip install -r requirements.txt + +# Configure credentials +cp .env.example .env +# Edit .env with your Fastn UCL API Key and Space ID + +# Run the sample +python app.py +``` + +## 🔌 Tool Coverage + +Fastn UCL supports tools across categories like: + +**Productivity** +- Gmail, Google Calendar, Notion, Slack, Microsoft Teams + +**Project Management** +- Jira, Linear, Asana, Trello + +**CRM & Sales** +- Salesforce, HubSpot, Zoho CRM + +**Enterprise IT** +- Active Directory, Okta, Datadog, Splunk + +**Data & Analytics** +- MongoDB, Elasticsearch, Tableau + +And **900+ more** — [see full list](https://docs.fastn.ai/) + +## 🏗️ Architecture Overview +fasnt architecture image + + +Fastn UCL acts as your intelligent, context-aware middleware, no manual wiring, no auth headaches. + +## 📚 Resources + +- [Fastn UCL Docs](https://docs.fastn.ai/) +- [MCP Spec](https://modelcontext.org/) +- [Integration Samples](https://github.com/fastn-ai/ucl-integration-samples) + +## 🆘 Support + +- Email: support@fastn.ai +- Join discussions: [GitHub Discussions](https://github.com/fastn-ai) +- Raise issues: [GitHub Issues](https://github.com/fastn-ai)