# 🛰️ EdgeClaw

> OpenClaw fork that routes tasks between local and cloud models

Canonical HTML: https://shelldex.com/projects/edgeclaw/
JSON: https://shelldex.com/api/projects/edgeclaw.json

## Overview

EdgeClaw can read project files, run shell commands and recall work from earlier sessions. Its router uses task complexity and privacy settings to decide which model handles a request.

**Highlight:** EdgeClaw can route a request to a local or hosted model according to privacy settings and task complexity.

## Editorial Take

This OpenClaw fork makes the model decision part of each request. You might keep a document summary on your machine and let a different task use a cloud provider. Redaction is a distinct option when routing work outside the machine.

## Best For

Someone who wants to keep selected material local while retaining a hosted model for other work.

## Tradeoffs

The routing policy is part of your privacy setup. Check how it handles a synthetic sensitive request before submitting real records.

## Metadata

- GitHub: https://github.com/OpenBMB/EdgeClaw
- Language: typescript
- Status: active
- Ecosystem role: fork
- Categories: self-hosted, privacy
- Platforms: cli, server
- Model options: Local option: Ollama or vLLM for locally routed requests. Hosted models may bill for usage.
- Added to registry: 2026-11-07

## GitHub data
Metrics from 2026-11-07T14:55:12.883Z.

- Stars: 1.2k (1,186)
- Forks: 70
- Open issues + pull requests: 1
- License: MIT
- Latest repository push: 2026-04-15T09:25:35Z

## Registry position

- Leaderboard rank: #29
- Health score: 40/100
- Days since last repository push at collection: 206

[Project setup instructions](https://github.com/OpenBMB/EdgeClaw#-quick-start).

Suggested first task: Point it at a spare project folder containing a short README. Ask it to identify the setup steps and quote the lines where it found them. Leave the files unchanged.

[How Shelldex measures metrics](https://shelldex.com/methodology/).
