I Built an Open-Source Cowork in 24 Hours with ZERO Rust Experience — and the Agent Builts Itself

Source: Dev.to

I walked into this with exactly zero Rust experience and a stubborn belief: building an agent shouldn’t require a megachurch of frameworks. Twenty‑four hours later, I shipped a Rust‑native AI Agent desktop for non‑devs that compiled to a 16 MB binary, a simple while loop with Skills, MCPs supported, BYOK (Bring Your Own Key) and even local models!

The Stack

• Rust – from zero to shipping in one day
• Long‑term memory – file system as persistent, queryable context
• Loop – a single while loop coordinating I/O and tool calls
• Tools – built‑in file tools (read, write, bash) + MCP‑compliant providers (extensible)
• Sandboxing – a simple Docker container isolates most risks

Secret Sauce

• Keep state in plain files and directories.
• Keep orchestration in a simple loop, not a framework.
• Teach the AI to read and write its own operating context, in files.

What I Learned

1. You don’t need an agent framework

Frameworks often add unnecessary abstraction and complexity. A simple loop suffices:

• Observe – read inputs, logs, and task files
• Decide – ask the model for the next atomic action
• Act – call a tool (MCP) or write to disk
• Reflect – append outcomes to a local journal

2. The file system is all you need

Vector stores and RAG farms are great, but they can become speculative overhead.

• Everything as files: tasks, plans, diffs, decisions, and logs live as plain text or JSON.
• Modern RL‑trained models work well with Linux file systems.
• Using the FS as “long‑term memory” makes persistence trivial and recoverability obvious.

3. The surreal mirror: AI coding reproduces itself

I used an AI agent (Claude Code) to help build the project. The Open Cowork is essentially a GUI with a Copilot‑like agent behind it, showing how an AI can reproduce itself in Rust.

4. Security as a first‑class citizen

Agents are powerful, but without boundaries they become liabilities. For non‑devs, security isn’t just a feature—it’s the foundation.

• Hard sandboxing (e.g., Docker) and strict whitelisting of system commands or network calls keep the agent helpful, not a system threat.

Code Overview

The entire codebase can be summarized in the loop below:

---

The project is open source:

Feel free to explore, try it out, and share any feedback!