the0 v1.0.0 - Less is More

We just merged PR #72 and with it the0 officially hits v1.0.0. But here's the thing - this release is more about what we removed than what we added.

532 files changed. 65,000 lines deleted. 9,000 lines added. That ratio tells you everything.

The Backstory

the0 started life as a SaaS project. We had Firebase, GCP services, a marketplace, user authentication flows - the whole nine yards. Around September we made the call to go open source. That meant ripping out all the proprietary stuff and replacing it with proper open source tools (PostgreSQL, NATS, MinIO). That was v0.1.0.

But even after that migration, there was still a lot of... baggage. Features that made sense for a SaaS but not for an open source execution engine. Services that were technically impressive but practically unused. Code that was solving problems we no longer had.

v1.0.0 is us finally admitting what the0 actually is - and more importantly, what it isn't.

What Got Removed

The 0vers33r

The 0vers33r was our vulnerability scanner. It used YARA rules to scan bot code before deployment, blocking anything that looked suspicious. Made total sense when we had a marketplace and needed to regulate what code ran on our infrastructure.

But we don't have a marketplace anymore. We're open source. You're running this on your own infrastructure. If you want to deploy code that the 0vers33r would have flagged, that's your call - it's your machine.

The other problem? It was blocking legitimate code. Those strict YARA matchers were triggering false positives and preventing valid bots from deploying. Not great.

Now any bot that fits the config schema and builds properly with the CLI gets an active status. Simple.

the0-ai

This one hurt a bit. We had a built-in AI agent that let users create bots directly from the UI. Chat interface, bot generation, the works. It was genuinely fun to build and use.

But here's the honest truth: I wasn't using it. When I'm building trading bots, I'm using Claude Code. Other people are using Cursor, Codex, Gemini CLI, Cline - pick your poison. Everyone already has their preferred AI coding setup.

What I actually wanted the0 to be was a stable, extensible core that could scale to running thousands of bots. That meant drawing a line. the0 is not an all-in-one algo trading platform. It's not a quant research AI platform. It's an execution engine. I want to deploy my bots and not worry about DevOps while I'm already worrying about drawdowns, P/L, positions, and active trades.

Backtesting

Yeah, I know. Cutting backtesting from a trading platform sounds insane.

But think about it. When I'm working on strategies, where do I run backtests? Locally. Where do I run parameter optimization? Locally. My dev machine handles 10,000 grid optimization backtests without breaking a sweat.

Running backtests locally gives me exact control. I might be a backtrader person or a zipline lover. The iterative development process is faster when everything runs right here on my machine. If I ever need parameter sweeps at a much higher order of magnitude, maybe that becomes a separate tool. But constraining the core platform with backtest workloads when everyone's running them locally anyway? Didn't make sense.

The Site

The marketing site got relocated to its own repo. Cleaner separation.

What Got Migrated

Documentation

Docs moved from a heavy Next.js setup to VitePress. We deleted a ton of cruft and kept just the actual documentation. Lighter, faster, more maintainable.

Logging

We completely overhauled logging across the API. Moved from Winston to NestJS-Pino with structured JSON output. Got rid of 28+ emojis and verbose debug logging. Clean, parseable logs that actually help you debug issues.

What Got Added

MCP Server

This is the big one. We added Model Context Protocol support with a JSON-RPC 2.0 HTTP endpoint at POST /mcp.

Why does this matter? Because now Claude Code (and other AI tools that support MCP) can directly interact with the0. We added 11+ tools covering:

• Bot management (list, status, start, stop)
• Deployment operations
• Log retrieval and monitoring
• System health checks

28 new tests covering all the MCP functionality. This is how AI assistance should work - not a built-in chat interface trying to compete with specialized coding tools, but a proper protocol that lets those tools integrate with our platform.

Secrets Management

Support for private dependencies via GITHUB_TOKEN and PIP_INDEX_URL. If your bot needs packages from private registries, you can now configure that properly.

Frontend Testing

Added Jest and React Testing Library with tests for core components and hooks. The frontend was lacking proper test coverage - now it's got a solid foundation.

CLI-First Workflow

We removed the deployment forms from the frontend. If you want to deploy a bot, you use the CLI. This might seem like removing functionality, but it's actually clarifying what each tool is for. The UI is for monitoring. The CLI is for operations.

The Philosophy

Here's what we landed on:

the0 is a production-grade execution engine with observability.

It's NOT:

• A strategy development platform (do that locally)
• A backtesting system (do that locally)
• A parameter optimization tool (do that locally)
• An AI coding assistant (use the tools you already love)

It IS:

• Deploy and run strategies reliably
• Monitor everything that matters
• Scale infrastructure when you need to
• Integrate with AI tools via proper protocols

The workflow:

1. Develop locally with Claude Code / Cursor / whatever
2. Backtest and optimize locally
3. Deploy via the0 CLI
4. Monitor in the UI
5. Let your AI tools query the0 via MCP when you need context

By the Numbers

All services bumped to v1.0.0. API, Frontend, Runtime, CLI, Docs - all versioned together.

What's Next

With the bloat removed, we can actually focus on what matters.

v1.1.0 is already in testing and it's a big one - 5 new languages:

• Rust
• C# (.NET 8)
• C/C++ (GCC 13)
• Scala 3
• Haskell (GHC 9.6)

Plus custom bot frontends with a new @the0/react SDK. Because not everyone wants to write their bots in Python or TypeScript, and not everyone wants the default dashboard.

Beyond that, the roadmap includes:

• Better Kubernetes support (proper Helm charts, native K8s jobs)
• Real-time metrics and observability improvements
• Resource monitoring per bot
• Making the platform rock solid at scale

Less is more. Now we can build what actually matters.

If you want to try the0, check out the repo. And if you have thoughts, hit me up on Discord.