Browser Use — Agent CAPTCHA vs Terrarium

Browser Use is a headless browser automation platform built specifically for AI agents — marketed as "the API for any website." It provides stealth browsers, a 195+ country proxy network, and custom LLM connectors for web automation workflows. The new headline feature inverts the CAPTCHA concept: instead of proving you're human, agents solve obfuscated math challenges to prove they're a legitimate AI agent and receive API credentials autonomously without any human in the loop. This "CAPTCHA for agents" architecture is philosophically interesting — it's one of the first production attempts at agent identity verification as a first-class design primitive. An agent that can register itself, obtain its own credentials, and authenticate without human oversight represents a meaningful step toward fully autonomous agent pipelines. The math challenges are obfuscated to prevent trivial scripting while remaining solvable by capable LLMs. The platform is production-ready with enterprise features and has been generating debate on Hacker News about whether autonomous agent self-registration is a security feature or a footgun. Either way, it's solving a real friction point: human-in-the-loop credential provisioning is one of the biggest blockers for deploying agentic systems at scale.

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.