Skip to content

SDK Overview

Flow-Like provides official SDKs for Node.js/TypeScript and Python that wrap the REST API into ergonomic, type-safe clients. Use them to control workflows, manage files, query LanceDB databases, run chat completions, generate embeddings, from your application code.

LanguagePackageInstall
Node.js / TypeScript@flow-like/sdknpm install @flow-like/sdk
Pythonflow-likeuv add flow-like
FeatureNode.jsPython
Trigger workflows (sync SSE + async)✅✅
Trigger events✅✅
File management (upload, download, list, delete)✅✅
LanceDB integration (credentials, queries, connections)✅✅
Chat Completions (+ streaming)✅✅
Responses (+ streaming)✅✅
Embeddings✅✅
Model discovery (LLMs, embeddings)✅✅
Board CRUD✅✅
App management✅✅
HTTP sinks✅✅
Execution monitoring / polling✅✅
LangChain integration✅✅
Async / await✅ native✅ a-prefixed methods

Model discovery returns api_surface alongside bit_id. Use that field when choosing the client method:

api_surfaceNode.jsPythonRequest shape
ChatCompletions or absentchatCompletions()chat_completions()messages, max_tokens
Responsesresponses()responses()input, max_output_tokens

These are separate request and response contracts. The server rejects a model sent to the wrong endpoint with HTTP 400 and names the required route. See the Node.js examples and Python examples for discovery and calls.

Both SDKs support two authentication methods:

TypePrefixHeader sentEnv variable
Personal Access Tokenpat_Authorization: pat_{id}.{secret}FLOW_LIKE_PAT
API Keyflk_X-API-Key: flk_{app}.{key}.{secret}FLOW_LIKE_API_KEY

The SDK sends PATs in Authorization and App API keys in X-API-Key. Use the explicit pat / apiKey options in Node.js, or pat / api_key in Python. Python also accepts a token option and detects its prefix. Provide one credential type at a time; Python rejects simultaneous PAT and API-key configuration.

Both SDKs read from environment variables by default:

Terminal window
export FLOW_LIKE_BASE_URL=https://api.flow-like.com
export FLOW_LIKE_PAT=pat_myid.mysecret
# or
export FLOW_LIKE_API_KEY=flk_appid.keyid.secret

Both SDKs include optional LangChain-compatible wrappers so you can use Flow-Like models inside LangChain chains, agents, and RAG pipelines.

  • Node.js: import { FlowLikeChatModel, FlowLikeEmbeddings } from "@flow-like/sdk/langchain"
  • Python: from flow_like.langchain import FlowLikeChatModel, FlowLikeEmbeddings

See the language-specific pages for installation and usage details.