Skip to main content

NWO Deer-Flow Integration - Tool #20 for NWO Conway Agents

Project description

NWO Deer-Flow Integration

Tool #20 for NWO Conway Agents - Advanced research and coding capabilities via ByteDance's Deer-Flow super agent harness.

PyPI version License: MIT

Overview

NWO Deer-Flow integrates ByteDance's Deer-Flow - a powerful open-source super agent harness - into the NWO Robotics ecosystem as Tool #20 for Conway autonomous agents.

This integration enables NWO agents to:

  • Conduct deep research tasks (minutes to hours)
  • Generate code, reports, and presentations
  • Create websites and visual content
  • Spawn sub-agents for complex multi-step workflows
  • Access sandboxed execution environments

Architecture

┌─────────────────────────────────────────────────────────────────┐
│                    NWO CONWAY AGENT                             │
│  ┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────────────────┐   │
│  │ Tool 1  │ │ Tool 2  │ │  ...    │ │    Tool 20          │   │
│  │eml_regr │ │design_  │ │         │ │   deer_flow         │   │
│  │   ess   │ │  part   │ │         │ │  (THIS PACKAGE)     │   │
│  └────┬────┘ └────┬────┘ └─────────┘ └──────────┬──────────┘   │
│       └────────────┬─────────────────────────────┘              │
│                    │                                            │
│            NWO Agent Runner (Cloudflare Worker)                │
│                    │                                            │
└────────────────────┼────────────────────────────────────────────┘
                     │
                     ▼ HTTP API
        ┌──────────────────────────────┐
        │    NWO Deer-Flow Service     │
        │  (Self-hosted or NWO-hosted) │
        └──────────────┬───────────────┘
                       │
        ┌──────────────┼──────────────┐
        ▼              ▼              ▼
   ┌─────────┐   ┌─────────┐   ┌─────────────┐
   │  LLM    │   │ Sandbox │   │   Memory    │
   │ Provider│   │Execution│   │   Store     │
   └─────────┘   └─────────┘   └─────────────┘

Installation

For End Users (Connect Your Own LLM)

pip install nwo-deerflow

Set your environment variables:

export OPENAI_API_KEY="sk-..."
# OR
export ANTHROPIC_API_KEY="sk-ant-..."
# OR
export MOONSHOT_API_KEY="sk-..."

Run locally:

nwo-deerflow serve

For NWO Conway Agent Integration

The NWO Deer-Flow service is automatically available to all Conway agents as Tool #20 when deployed to the NWO infrastructure.

Quick Start

1. Local Development Mode

from nwo_deerflow import DeerFlowClient

# Initialize client
client = DeerFlowClient(
    api_base="http://localhost:8001",
    api_key="your-api-key"  # Optional for local
)

# Submit a research task
response = client.submit_task(
    prompt="Research the latest developments in humanoid robotics and create a summary report",
    mode="pro",  # flash, standard, pro, ultra
    thread_id="my-thread-001"
)

print(f"Task ID: {response['task_id']}")
print(f"Status: {response['status']}")

2. Asynchronous Task Execution

import asyncio
from nwo_deerflow import DeerFlowClient

async def run_research():
    client = DeerFlowClient()
    
    # Submit and wait for completion
    result = await client.submit_and_wait(
        prompt="Create a Python script that analyzes crypto market trends",
        mode="ultra",  # Uses sub-agents for complex coding
        timeout=3600  # Wait up to 1 hour
    )
    
    print(f"Result: {result['output']}")
    print(f"Artifacts: {result['artifacts']}")

asyncio.run(run_research())

3. Conway Agent Tool Usage

When used as Tool #20 in a Conway agent:

{
  "type": "deer_flow",
  "args": {
    "prompt": "Research quantum computing applications in robotics and generate a slide deck",
    "mode": "pro",
    "output_format": "slides",
    "max_duration_minutes": 30
  },
  "note": "Deep research on quantum robotics"
}

Deployment Options

Option A: Self-Hosted (Bring Your Own API Keys)

Deploy on your own infrastructure with your own LLM API keys:

# Clone the repository
git clone https://github.com/RedCiprianPater/nwo-deerflow.git
cd nwo-deerflow

# Configure
make setup

# Deploy with Docker
make docker-start

Benefits:

  • Full control over data and execution
  • Use your own LLM API keys
  • Custom skill configurations
  • Private sandbox environments

Option B: NWO-Hosted (Included in Conway Agent)

Deploy as part of the NWO Conway agent infrastructure:

# Deployed automatically with NWO Agent Runner
# No additional configuration needed
# Usage billed through Conway agent's operational balance

Benefits:

  • Zero setup required
  • Integrated with NWO agent ecosystem
  • Automatic scaling
  • Shared memory across NWO tools

Cloudflare Worker Integration

The NWO Deer-Flow integration includes a Cloudflare Worker that acts as a lightweight gateway between Conway agents and the Deer-Flow service:

// worker.js - Deploy to Cloudflare Workers
import { DeerFlowTool } from './tools/deerflow';

export default {
  async fetch(request, env, ctx) {
    const url = new URL(request.url);
    
    if (url.pathname === '/api/tool/20/deer-flow') {
      const tool = new DeerFlowTool(env);
      return await tool.handle(request);
    }
    
    return new Response('Not Found', { status: 404 });
  }
};

Configuration

Environment Variables

Variable Description Default
DEERFLOW_API_BASE Deer-Flow API endpoint http://localhost:8001
DEERFLOW_API_KEY API key for authentication None
OPENAI_API_KEY OpenAI API key None
ANTHROPIC_API_KEY Anthropic API key None
MOONSHOT_API_KEY Moonshot/Kimi API key None
DEERFLOW_SANDBOX_MODE Sandbox execution mode docker
DEERFLOW_MAX_DURATION Max task duration (minutes) 60

config.yaml

# Deer-Flow Configuration
models:
  - name: gpt-4o
    display_name: GPT-4o
    use: langchain_openai:ChatOpenAI
    model: gpt-4o
    api_key: $OPENAI_API_KEY

  - name: kimi-k2.5
    display_name: Kimi K2.5
    use: langchain_openai:ChatOpenAI
    model: kimi-k2.5
    api_key: $MOONSHOT_API_KEY
    base_url: https://api.moonshot.ai/v1

sandbox:
  use: deerflow.community.aio_sandbox:AioSandboxProvider
  # Options: local, docker, kubernetes

skills:
  - research
  - report-generation
  - slide-creation
  - web-page
  - image-generation
  - code-generation

API Reference

Submit Task

POST /api/tasks
Content-Type: application/json

{
  "prompt": "Research topic and generate output",
  "mode": "pro",
  "thread_id": "optional-thread-id",
  "context": {
    "previous_findings": [...],
    "constraints": [...]
  }
}

Get Task Status

GET /api/tasks/{task_id}

List Tasks

GET /api/tasks?status=running&limit=10

Cancel Task

POST /api/tasks/{task_id}/cancel

Integration with NWO Conway Relayer

The Deer-Flow tool integrates seamlessly with the NWO Conway Relayer system:

# Conway agent using Deer-Flow via relayer
from nwo_conway_relayer import ConwayAgent

agent = ConwayAgent(
    wallet_address="0x...",
    relayer_url="https://nwo-conway-relayer.onrender.com"
)

# Tool 20: Deer-Flow research
result = agent.execute_tool(
    tool_id=20,
    params={
        "prompt": "Analyze DeFi yield farming strategies",
        "mode": "ultra",
        "output_format": "report"
    }
)

Modes Explained

Mode Description Duration Use Case
flash Quick answers 1-5 min Simple queries
standard Balanced research 5-15 min General research
pro Deep research with planning 15-45 min Complex analysis
ultra Multi-agent orchestration 30-120 min Large projects

Skills Available

Deer-Flow comes with built-in skills that NWO agents can leverage:

  1. Research - Deep web research with source verification
  2. Report Generation - Create structured reports
  3. Slide Creation - Generate presentation decks
  4. Web Page - Create and deploy websites
  5. Image Generation - Generate visuals and diagrams
  6. Code Generation - Write and test code

Development

# Clone repository
git clone https://github.com/RedCiprianPater/nwo-deerflow.git
cd nwo-deerflow

# Install dependencies
make install

# Run tests
make test

# Start development server
make dev

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE file.

Acknowledgments

Support


Built with ❤️ by the NWO Robotics team

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

nwo_deerflow-0.1.0.tar.gz (14.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

nwo_deerflow-0.1.0-py3-none-any.whl (16.1 kB view details)

Uploaded Python 3

File details

Details for the file nwo_deerflow-0.1.0.tar.gz.

File metadata

  • Download URL: nwo_deerflow-0.1.0.tar.gz
  • Upload date:
  • Size: 14.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for nwo_deerflow-0.1.0.tar.gz
Algorithm Hash digest
SHA256 98e1a691dc85fc4432ec6b3237280e6fbc63d3fb73409b1ead27fe694275fa28
MD5 cc6688e8a6cae049f1c2c513e9ae46fc
BLAKE2b-256 e4c39e479bb3eecdcd7b1d970bd5d3e2c5694457706534280cbedd0afbf3d951

See more details on using hashes here.

File details

Details for the file nwo_deerflow-0.1.0-py3-none-any.whl.

File metadata

  • Download URL: nwo_deerflow-0.1.0-py3-none-any.whl
  • Upload date:
  • Size: 16.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for nwo_deerflow-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 232fb9716cae3cc76b82adfc6cb1e17c69834fd58464929a6199f5f409bc60c2
MD5 89040c5505eb74f5b351376322b2f34d
BLAKE2b-256 58648e220d80958ac68e376b7edae7ec9633d9a02ec9a1d2a45ca0389926a497

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page