Claw Code vs Terrarium

On March 31, 2026, a security researcher discovered that Anthropic had accidentally published full Claude Code source maps to npm — making the entire internal architecture readable to anyone who looked. Within hours, a developer going by ultraworkers began a clean-room rewrite in Rust, and Claw Code was born. The project hit 180,000 GitHub stars in under two weeks, making it one of the fastest-growing open-source repositories in history. It replicates Claude Code's core agent loop, permission system, and tool dispatch while adding a Rust-native performance profile and removing telemetry. The project explicitly operates under clean-room principles — contributors who viewed the source maps are excluded from contributing. The implications are significant: Claw Code is proof that the underlying architecture of agentic coding tools is now commoditized. If Anthropic's secret sauce was the agent loop, that loop is now public. What remains is the model quality — and Claw Code works with any API-compatible provider.

Terrarium is a multi-turn evaluation and optimization engine for LLM agents built by evolvent-ai. Unlike static benchmark suites that measure agents against fixed input-output pairs, Terrarium creates persistent, stateful "living environments" — simulated deployment contexts where agents operate over extended sessions, accumulate state, use tools, and interact with simulated external systems. You evaluate agents the way you'd test a car: by driving it, not by measuring its doors. The system supports configurable environment complexity, including simulated databases, APIs, file systems, and user personas. Agents are scored not just on final outputs but on trajectory quality — how efficiently they reached the answer, how often they hallucinated intermediate steps, and how well they recovered from dead ends. The engine also supports continuous optimization loops where poor-performing trajectories trigger automatic prompt refinement. With 17 stars and created April 14, Terrarium is extremely new. But it's addressing a genuine gap: the disconnect between how agents perform on static benchmarks versus how they behave in production. As enterprise AI deployments scale, the need for realistic pre-production evaluation is becoming critical.