LocalForge

Local-first AI development platform — VS Code extension · CLI · Web UI
Multi-agent workflow: Planner → Writer → Reviewer → Tester
Proudly Made in Canada. 🇨🇦

Features • Quick Start • Modes • Architecture • Pricing • License

---

Features

Capability	LocalForge	Copilot	Cursor	Claude Code
Fully offline	✅	❌	❌	❌
Any LLM provider (75+)	✅	❌ (1)	⚠️ (limited)	❌ (1)
Multi-agent workflow	✅	❌	❌	❌
CLI + Web UI	✅	❌	❌	✅
Self-hosted server	✅	❌	❌	❌
Zero data leaving your machine	✅	❌	❌	❌
AES-256-GCM encryption	✅	❌	❌	❌
Price	$0-$29/mo	$10-39/mo	$20/mo	$20/mo

Quick Start

Prerequisites

• Node.js 20+
• For local models: llama.cpp server + GGUF file
• For cloud models: opencode CLI (npm i -g opencode) or an OpenAI API key

VS Code Extension

npm install
npm run build
# Press F5 in VS Code → click "LocalForge" in status bar

CLI

# Interactive
npm start

# One-shot
npm run run "add input validation to my Express API"

# Multi-agent workflow (the killer feature)
npm run workflow "build a REST API with CRUD endpoints"

# Plan mode
npm run plan "design auth system architecture"

# Demo mode (no LLM needed)
$env:LOCALFORGE_DEMO = "1"; npm start

Web UI

npm run server
# Open http://127.0.0.1:3096

# With API key auth
$env:LOCALFORGE_API_KEY = "your-secret"; npm run server

Providers

Variable	Default	Description
LOCALFORGE_PROVIDER	local	demo, local, opencode, openai
LOCALFORGE_ENDPOINT	http://127.0.0.1:11434/v1	API endpoint
LOCALFORGE_MODEL	qwen2.5-coder-7b-instruct-q4_k_m	Model name
LOCALFORGE_API_KEY	—	API key for auth

Modes

Mode	Description
Chat	Conversational AI with code context and optional mode switching (@plan, @build)
Agent	Autonomous: plans, executes, iterates up to 5 rounds with task feedback
Plan	Architecture analysis with structured JSON output and mode recommendations
Build	Multi-agent workflow: Planner → Writer → Reviewer → Tester pipeline

Architecture

src/
├── extension.ts            VS Code extension (webview sidebar)
├── cli.ts                  CLI entry point (stdin/stdout, one-shot)
├── server.ts               Standalone web UI server (port 3096)
├── AgentTask.ts            Core engine: modes, conversation, agent loop
├── Workflow.ts             Multi-agent workflow pipeline
├── Mode.ts                 Mode definitions and system prompts
├── providers/
│   ├── ProviderManager.ts  LLM abstraction: demo/local/opencode/openai
│   └── DemoProvider.ts     Zero-dependency demo responses
└── sandbox/
    └── Executor.ts         File write, terminal exec, test runner, git auto-commit

Security

• Zero-knowledge architecture — no LocalForge cloud server
• AES-256-GCM encryption for conversations (optional)
• Command approval — granular control over file/terminal/test operations
• Destructive command blocklist — prevents rm -rf / and similar
• Binary file protection — blocks arbitrary binary writes
• Configurable via SecurityConfig interface (encryptConversations, commandApproval)

Pricing

Tier	Price	Best for
Self-Hosted	$0 forever	Open source, hobbyists, offline use
Pro	$14/mo	Professional developers needing 75+ providers
Enterprise	$29/mo	Teams needing SSO, private deployment, support
Perpetual	$249 one-time	Air-gapped environments, 1 year updates

Contributing

Pull requests welcome! See CONTRIBUTING.md for details.

• master is protected — all PRs need review + passing CI
• Use issue templates for bugs and feature requests
• Questions? Start a GitHub Discussion

License

Copyright (C) 2026 Nrupal Akolkar. Proudly Made in Canada.

The LocalForge source code is licensed under the GNU Affero General Public License v3 with a Commercial License Exception.

• Open source use (personal, non-commercial, or AGPL-compatible): Free under AGPL v3
• Commercial use (for-profit entity, proprietary software): Requires a commercial license
• See LICENSE for full terms

Commercial licenses available at localforge.dev/pricing.
Support on Buy Me a Coffee ☕ — Payments via Stripe.

---

Proudly Made in Canada 🇨🇦