Local and cloud LLM support
Use LM Studio on your machine, remote OpenAI-compatible endpoints, OpenRouter, or custom providers.
Local-first AI software forge
Prompt it. Watch it plan, build, test, debug, and package.
ByteForge is a Windows-first AI development workspace for people who want software made from a prompt without surrendering their machine, files, or model choices.
LM Studio ready
Tailscale endpoints
Cloud compatible
Workspace safe
4+Provider types
8Starter templates
100%Local workspace control
Plan -> ReleaseTarget workflow
What makes it different
Not a chatbot. A controlled development workstation.
ByteForge is being built around the practical loop developers actually need: understand the request, inspect the workspace, choose the right tools, create code, run checks, debug failures, and stop only when the project is ready to test or release.
Model capability checks
ByteForge probes selected models and adapts tool strategy instead of assuming every local model speaks the same format.
Prompt-inferred templates
Generate web apps, static sites, Python tools, Node APIs, Android starters, .NET desktop apps, and games.
Build, test, debug, fix
Approved commands run inside the workspace, errors are surfaced, and the agent can continue with targeted fixes.
Workflow
The intended project lifecycle.
Prompt
Describe the app, site, script, API, or tool in plain language.
Plan
ByteForge infers platform, files, dependencies, and verification steps.
Build
The agent creates or patches files using the safest model-compatible tool path.
Test
It runs install/build/test commands when approved or trusted mode is enabled.
Debug
Failures become context: read error, inspect file, patch, rerun.
Release
The final output is ready for preview, packaging, or handoff.
Model freedom
Use the model stack you actually want.
ByteForge is designed for local-first development with LM Studio, while still supporting OpenAI-compatible remote endpoints. The goal is simple: pick a model, verify it can operate tools safely, then let the agent work through the project.
LM Studio
OpenAI-compatible
OpenRouter
Custom endpoint
Qwen / GGUF harness
Reasoning levels
Early Windows build
Download, test, and shape the product.
ByteForge is in active development. The current build is for early adopters, workflow testing, model compatibility checks, and project-generation trials.