Show HN：Orcheo – 一款類似 n8n 的 Python 工作流引擎，專為 AI 代理而建

Navigation Menu

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

To see all available qualifiers, see our documentation.

A full-stack, agentic workflow programming platform. Built for vibe-coding and with vibe-coding.

License

Uh oh!

There was an error while loading. Please reload this page.

ShaojieJiang/orcheo

Folders and files

Latest commit

History

Repository files navigation

Orcheo

Orcheo is a workflow orchestration platform designed for vibe coding — AI coding agents like Claude Code can start services, build workflows, and deploy them for you automatically.

Note: This project is currently in Alpha. Remote development and deployment are still to be tested.

Why Orcheo?

For users

Quick start

The project ships with everything needed to spin up the FastAPI runtime on
SQLite for local development.

Install dependencies

For development (from source):

Or install from PyPI:

Activate the virtual environment (optional but recommended)

Once activated, you can run commands without the uv run prefix.

Run the API server

If you enable authentication, configure a bootstrap token first to create your initial service tokens:

Bootstrap tokens are special environment-based service tokens for initial setup only:

After the server starts, use the bootstrap token to create persistent service tokens:

Then switch to the new persistent token and remove the bootstrap token:

Verify the setup

CLI

Orcheo ships with a LangGraph-friendly CLI for node discovery, workflow
inspection, credential management, and reference code generation.

Getting Started

After activating the virtual environment, get started with:

Shell Auto-Completion

Enable fast shell auto-completion for commands and options:

This installs completion for your current shell (bash, zsh, fish, or PowerShell).
After installation, restart your shell or source your shell configuration file.

Published workflows remain accessible until you run orcheo workflow unpublish <workflow> or toggle the --require-login flag to gate public chats behind OAuth.

Pass workflow inputs inline with --inputs or from disk via --inputs-file. Use --config or --config-file to provide LangChain runnable configuration for the execution (each pair is mutually exclusive).

Upload-time defaults can be stored on a workflow version with orcheo workflow upload ... --config or --config-file. Stored config is merged with per-run overrides (run config wins). Avoid putting secrets in runnable config; use environment variables or credential vaults instead.

Pass --offline to reuse cached metadata when disconnected:

Authentication

Orcheo supports flexible authentication to protect your workflows and API endpoints.

Configure via the ORCHEO_AUTH_MODE environment variable:

Service tokens are ideal for CLI usage, CI/CD pipelines, and automated integrations.

Create a token:

Use with CLI:

Use with Python SDK:

Rotate tokens:

For production deployments with Identity Providers (Auth0, Okta, Keycloak):

For detailed configuration, security best practices, and troubleshooting, see docs/authentication_guide.md.

MCP (Model Context Protocol)

Orcheo SDK includes an MCP server that allows AI assistants like Claude to interact with your workflows.

To configure it in Claude Desktop, add the following to your claude_desktop_config.json:

Note: This configuration requires the Orcheo development backend to be running locally (see Run the API server).

To configure the MCP server in Claude CLI:

Note: Replace /path/to/uvx with your actual uvx binary path (find it with which uvx).

To configure the MCP server in Codex CLI:

Note: Replace /path/to/uvx with your actual uvx binary path (find it with which uvx).

Canvas (Visual Workflow Designer)

Orcheo Canvas is the visual workflow designer for creating, visualizing, and managing workflows.

After installation, start the Canvas interface:

The Canvas application will be available at http://localhost:5173 (dev mode) or the configured preview port (production mode).

For more details, see apps/canvas/README.md.

For developers

Repository layout

Opening the repository inside VS Code automatically offers to start the included
dev container with uv and Node.js preinstalled. The new quickstart flows in
examples/quickstart/ demonstrate the visual designer and SDK user journeys,
and examples/ingest_langgraph.py shows how to push a Python LangGraph script
directly to the backend importer, execute it, and stream live updates.

See docs/deployment.md for Docker Compose and managed
PostgreSQL deployment recipes.

Seed environment variables

To set up your development environment:

Pass --force to overwrite an existing .env file.

Configuration

The CLI reads configuration from:

See docs/cli_tool_design.md for detailed design,
roadmap, and future MCP server integration plans.

Custom nodes and tools

Learn how to extend Orcheo with your own nodes, tool integrations, and workflow helpers in docs/custom_nodes_and_tools.md.

Workflow repository configuration

The FastAPI backend now supports pluggable workflow repositories so local
development can persist state without depending on Postgres. By default the app
uses a SQLite database located at ~/.orcheo/workflows.sqlite. Adjust the
following environment variables to switch behaviour:

Refer to .env.example for sample values and to docs/deployment.md for
deployment-specific guidance.

About

A full-stack, agentic workflow programming platform. Built for vibe-coding and with vibe-coding.

Topics

Resources

License

Uh oh!

There was an error while loading. Please reload this page.

Stars

Watchers

Forks

Releases

Packages

  0

Uh oh!

There was an error while loading. Please reload this page.

Contributors

  2

Languages

Footer

Footer navigation