Flowiseai

Flowise is an open-source low-code platform for LLM application development — primarily a drag-and-drop flow builder that exports deployable REST endpoints. For agents: Flowise can host RAG pipelines, tool-using agents, and chatbot flows that other agents call via API. Also valuable as a visual development environment where product teams prototype agent flows before writing code. Self-hostable on any Docker-compatible infrastructure. Confidence is docs-derived.

REST API: POST /api/v1/prediction/{chatflowId} runs a prediction against a deployed chatflow ({question, sessionId, overrideConfig}). GET /api/v1/chatflows lists deployed chatflows. GET /api/v1/chatflows/{id} retrieves chatflow metadata. Streaming: POST /api/v1/prediction/{id} with stream=true for SSE streaming response. Chatflow definition via visual UI or JSON export.

API key auth for protected chatflows: Authorization: Bearer {api_key} header. Keys generated per chatflow (optional — flows can be public). HTTPS enforced. Self-hosted: own auth infrastructure. Flowise Cloud: platform-level auth. No OAuth2.

Chatflow execution: synchronous (prediction completes in response) or streaming (SSE). Execution errors: model provider errors surface in response body. Self-hosted reliability depends on Docker/infrastructure setup. Flowise Cloud uptime tracked separately. LLM provider downtime propagates to chatflow errors — implement fallback providers in critical flows.

docs.flowiseai.com covers quickstart (Docker/npm), node library reference, API usage, and integration guides. Getting started: docker run or npm start, access UI at localhost:3000, build first chatflow in under 30 minutes. Open-source MIT license (GitHub: FlowiseAI/Flowise, 20k+ stars). Community via Flowise Discord and GitHub. Actively maintained.