Introduction

​

What is Flowra?

Flowra is an AI-powered platform for building agents, workflows, and integrations. In the product, everything happens in chat: you build and edit workflows, create tools, add knowledge (RAG), manage data, and run automations from the same place — by talking to the assistant. You can also use the API to integrate with your apps: list and execute tools, manage auth and connected accounts, trigger workflows, and more. No need to wire each integration yourself: configure auth once, connect user accounts via OAuth, and run workflows on a schedule or from your app. Flowra gives you:

• 20,000+ tools — From any major API to your own custom tools. You can turn code (JavaScript or Python) into tools: test a script in the sandbox, then create a tool so agents and workflows can call it. Tools are grouped into toolkits (e.g. Slack, Gmail, Airtable). Search by keyword, filter by toolkit or tags, and execute with one API call.
• Workflows & agents — Orchestrate steps, run automations, and build AI agents that use these tools. In the product you create and edit workflows in chat; via the API you create, run, and inspect executions.
• Triggers — Run workflows on a schedule (cron) or when events occur. Create and manage trigger instances via the API so your automations fire without manual calls.
• Auth & connections — Auth configs (OAuth or API keys) for each toolkit, and connected accounts so end users can link their Slack, Google, etc. via a link flow. Tools run with the right credentials automatically.
• MCP — Model Context Protocol: register MCP servers, expose your project’s tools to MCP clients (Cursor, Claude Desktop, etc.), and call external MCP servers from your agents.
• Knowledge / RAG — Add files and URLs for retrieval-augmented generation so your agents can use your data.
• Files & data — Upload and list files; store and query structured data in collections (Table API). Use them in workflows, knowledge bases, or your own app logic.

You manage tools, auth configs, triggers, connected accounts, workflows, and data via the Flowra API. In the Flowra product, you do it all in chat: build workflows, create tools, add knowledge (RAG), manage data, and run everything from one place. Everything is scoped to a project and secured with API keys.

​

Workflow vs agent — what’s the difference?

In Flowra you’ll see both words; here’s the simple difference:

• Workflow — A defined sequence of steps: do step A, then B, then C. It can include tools (e.g. send Slack, call an API), conditions, and loops, but the path is largely fixed. You run it on a schedule (e.g. daily report) or trigger it manually. Good for repeating, predictable automations: “every morning pull data, transform it, post to Slack.”
• Agent — AI-driven: an AI model decides what to do next. It can use the same tools, but it reacts to the user’s messages or the situation — chat, answer questions, choose which tool to call, or ask for more info. Good for conversation and open-ended tasks: “talk to the user, understand the goal, then use tools as needed.”

In Flowra, a single thing can be both: you build one flow that has fixed steps in some places and AI (agent) steps in others. So “workflow” often means “the thing you build and run” (whether it’s more fixed or more AI). When we say “agent,” we usually mean the part that uses the AI to chat and decide. Either way, you build and run it in chat and via the API; the name just tells you whether the behavior is more fixed steps (workflow) or AI-decides (agent).

​

Who is it for?

Everyone who needs automation and AI agents — whether you code or not.

• No-code / business users — Do everything in chat in the Flowra product: build and edit workflows, create tools, add knowledge (RAG), manage data, connect accounts, set triggers, and run automations. No programming required; you talk to the assistant to create flows, tools, and data and to run them.
• Developers and teams — Use the API to list tools, create custom tools from code, manage auth and connected accounts, trigger workflows from your app, and integrate with MCP (Cursor, Claude Desktop). One platform for auth, tools, triggers, and execution instead of stitching many services.
• Both — Teams often combine both: non-developers build workflows, tools, RAG, and data in chat; developers extend and automate via the API and embed workflows in products.

​

What you can build

• Customer support bots — Connect Slack/email, add knowledge (files/URLs), and run agents that answer questions and create tickets.
• Scheduled reports — Triggers that run workflows daily or weekly to pull data from APIs, transform it, and post to Slack or email.
• Multi-tenant SaaS — External users per tenant; connected accounts so each user links their own Google/Slack; workflows that act on their behalf.
• Chat and assistants — Threads API for conversation sessions; stream agent runs via SSE; use the LLM endpoint for one-off completions.
• Internal tools — Table API for app data; file upload for attachments; workflows that move data between systems on a schedule or webhook. Turn your own code (JS/Python) into custom tools and call them from workflows like any other tool.
• MCP-powered IDE — Register MCP servers, expose your tools to Cursor or Claude so developers use them inside the editor.

​

How it works at a high level

1. Create a project and get an API key — In the Flowra Dashboard, create or select a project and create an API key under Project → API Keys. Use the key in the x-api-key header for all API calls. See Quickstart and API keys.
2. Configure auth — Create auth configs (OAuth or API keys) for the toolkits you need (e.g. Slack, Google). This tells Flowra how to authenticate with each service.
3. Connect user accounts — Use connected accounts: create a link with POST /api/v1/connected_accounts/link, send users to the returned URL so they complete OAuth, then list connected accounts and use them in workflows.
4. Discover and use tools — List and search tools via GET /api/v1/tools (keyword, toolkit, tags). Execute a tool with POST /api/v1/tools/execute/{toolSlug} and the right inputs. Optionally group tools in toolkits and attach them to workflows.
5. Create and run workflows — In the product you build and edit workflows, tools, knowledge (RAG), and data in chat by talking to the assistant. Via the API you list, create, run, and inspect executions. You can also trigger workflows automatically on a schedule or event.
6. Optional: chat API, data, MCP — Use threads for conversation sessions and streaming agent runs. Use files and Table for uploads and structured data. Configure an MCP server so Cursor, Claude, or other MCP clients can call your project’s tools.

The API covers workflows, tools, toolkits, auth configs, trigger instances, connected accounts, files, table (collections/documents), threads (Graphify), AI models, LLM generation, and MCP. In the product, you do it all in chat: build workflows, create tools, add RAG, manage data, and run everything from one place.

​

Capabilities

​

Quick start

Get your API key from the dashboard and make your first request in a few minutes. Then explore the guides for the areas you need.

​

API reference

All API endpoints are documented from the OpenAPI spec. You can try requests from the docs and see request/response schemas for every endpoint.