Deploy autonomous coding agents on your local hardware.
Use your own API keys. Zero cloud markups. Absolute code privacy.
Orchestrate Claude, Gemini, Codex, and local LLMs from anywhere.
Your code never leaves your machine. CallByte is the orchestration layer — not a host.
Deploy a single Bare Metal agent — Claude, Gemini, or local models — into any local Git repository. Chat directly from your phone via Telegram, or use our dedicated developer chat interface to review code and debug in real-time.
Fix the auth middleware — JWT tokens expire but the refresh loop isn't catching 401s
Found it. The interceptor catches 401 but doesn't await the refresh before retrying:
+ await refreshToken();
+ return api.request(config);
- return Promise.reject(err);
const limiter = rateLimit({
windowMs: 15 * 60 * 1000,
max: 100, store: redisStore,
});
Why rely on one model? Spin up Claude Sonnet to write the feature, and drop Gemini 3.1 Pro into the same directory to audit the security. Hot-swap models on the fly and watch them collaborate in the same Git Worktree.
Stop micromanaging prompts. Deploy a CEO Orchestrator that analyzes your codebase, writes Jira-style tickets, and autonomously spawns specialized worker agents — Frontend, Backend, QA — to execute the work in parallel isolation.
Add auth guard to /settings route
T-003
Implement WebSocket rate limiter
T-001 · cb-worker-backend
E2E test: auth + rate limit flow
T-002 · cb-worker-qa
Add JWT refresh endpoint
T-000 · shipped
You bring your own API keys and hardware. We charge for orchestration — not compute.
For trying things out
For power users
For dev teams & agencies
For organizations
Install the agent on your machine, connect to CallByte, and start shipping with autonomous AI agents in under 5 minutes.
Get Started