neuron-explorer
Steering-first neuron explorer for Qwen3.5-4B: push every MLP neuron, keep every rollout, browse the results
Agent environments that observe, measure, and understand themselves
Developer toolsTypeScript · JavaScriptGitHub @The-Focus-AIOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
CLI tool for evaluating and comparing AI models across Google, Ollama, OpenRouter, LM Studio, LlamaBarn, and GitHub Models. Features robust error handling, cost tracking, memory-augmented chat, and dynamic test coverage.
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
Every AI model lives in its own perceptual bubble — its Umwelt. Umwelten lets you build agent environments that observe, measure, and understand themselves.
Habitat — a living container for AI agents: persona, tools, memory, sessions, sub-agents, multiple interfaces (CLI, Telegram, Discord, web). Evaluation — systematic model assessment that reveals how models actually see the world. TypeScript / Node 20+.Read the full README ↗Every AI model lives in its own perceptual bubble — its Umwelt. Umwelten lets you build agent environments that observe, measure, and understand themselves.
Habitat — a living container for AI agents: persona, tools, memory, sessions, sub-agents, multiple interfaces (CLI, Telegram, Discord, web). Evaluation — systematic model assessment that reveals how models actually see the world. TypeScript / Node 20+.
git clone https://github.com/The-Focus-AI/umwelten.git
cd umwelten && pnpm install && cp env.template .env
# Add your GOOGLE_GENERATIVE_AI_API_KEY to .env
# `umwelten search` requires ripgrep (rg) on PATH: brew install ripgrep
# Start an agent environment
npx umwelten habitat
# Run one prompt through a provider/model
npx umwelten run "Explain why the sky is blue" \
--provider google --model gemini-3-flash-preview
# 76% of models fail this common-sense question
dotenvx run -- pnpm tsx examples/evals/car-wash.ts
# Watch models fall for classic logic traps
dotenvx run -- pnpm tsx examples/evals/reasoning.ts
# Can a model write exactly 12 words?
dotenvx run -- pnpm tsx examples/evals/instruction.ts
import { Stimulus, Interaction, Habitat, EvalSuite } from "umwelten";
// Talk to any model
const stimulus = new Stimulus({ role: "helpful assistant" });
const interaction = new Interaction(
{ name: "gemini-3-flash-preview", provider: "google" },
stimulus,
);
interaction.addMessage({ role: "user", content: "Hello" });
const reply = await interaction.generateText();
// Build an agent environment
const habitat = await Habitat.create({ workDir: "./my-agent" });
const { interaction: ix } = await habitat.createInteraction();
ix.addMessage({ role: "user", content: "List my agents" });
await ix.generateText();
// Evaluate models
const suite = new EvalSuite({
name: 'quick-test',
stimulus: { role: 'helpful assistant', temperature: 0.3 },
models: [{ name: 'gemini-3-flash-preview', provider: 'google' }],
tasks: [{
id: 'math', prompt: 'What is 2+2?', maxScore: 1,
verify: (r) => ({ score: r.trim() === '4' ? 1 : 0, details: r.trim() }),
}],
});
await suite.run();
Habitats also drive HTTP chat. @umwelten/habitat exposes startWebServer — the web peer to the Discord and Telegram adapters. It speaks the Vercel AI SDK UI Message Stream Protocol, so any React frontend using @ai-sdk/react's useChat connects with no glue code. Streaming text, tool calls, and tool results all flow through the same ChannelBridge that every other channel uses.
import { Habitat } from 'umwelten';
import { startWebServer } from '@umwelten/habitat';
const habitat = await Habitat.create({ workDir: './my-habitat' });
await startWebServer({
habitat,
auth: 'dev', // or a custom AuthProvider
staticRoot: './public', // built SPA
port: 3000,
});
Generative UI comes via a renderUi tool that ships a json-render Spec through the tool-call channel — the client renders it with @json-render/react. See examples/umwelten-web-demo/ for the reference app (useChat + thread sidebar + tool-call cards + renderUi).
mise run web-demo # API server on :3000
mise run web-demo-client # Vite dev server on :5173 (proxies /api)
Gaia — the built-in habitat manager UI — is a thin wrapper over the same framework (mise run habitat-web).
Connect to any remote MCP server with OAuth — chat from the command line or build your own MCP-backed agent.
# Chat with a remote MCP server (OAuth handled automatically)
npx umwelten mcp chat --url https://oura-mcp.fly.dev/mcp
# One-shot query
npx umwelten mcp chat --url https://oura-mcp.fly.dev/mcp --one-shot "how did I sleep?"
Build your own multi-user MCP server with Habitat: see examples/oura-mcp/ — a complete Oura Ring MCP server with OAuth, Neon Postgres, deployable on fly.io.
Your Habitat reads Claude Code and Cursor history — every session, every tool call, every solution. Digest with AI, search full text, extract learnings.
npx umwelten sessions list # see your sessions
npx umwelten sessions index # AI-index everything
npx umwelten sessions search "authentication" # full-content search
npx umwelten sessions browse # interactive browser
Machine-oriented summary for agents: LLM.txt.
Monorepo managed with pnpm workspaces. Each package is independently publishable under the @umwelten/ scope.
packages/
core/ @umwelten/core — model runners, stimulus, interaction, providers, context, memory
protocols/ @umwelten/protocols — MCP, A2A, OAuth, protocol task storage
substrate/ @umwelten/substrate — reversible Components, Services, Shell composition
habitat/ @umwelten/habitat — agent container, tools, Gaia, web/A2A/MCP server
sessions/ @umwelten/sessions — session search, browse, digest commands
evaluation/ @umwelten/evaluation — EvalSuite, ranking, aggregation, reporting
ui/ @umwelten/ui — Telegram, Discord, TUI adapters
mycel/ @umwelten/mycel — the Exchange: dispatch, metering, balances
supplier/ @umwelten/supplier — supplier discovery, probing, publishing, serving
cli/ @umwelten/cli — umwelten CLI commands
umwelten/ umwelten — published compatibility/meta-package
examples/
evals/ — EvalSuite examples (car-wash, reasoning, instruction)
model-showdown/ — multi-dimension eval suite
oura-mcp/ — multi-user Oura Ring MCP server (Habitat + fly.io)
umwelten-web-demo/ — React + useChat reference app
habitat-minimal/ — minimal Habitat work-dir
See the current architecture map for runtime flows, the Gaia/SaaS boundary, old/new paths, and cleanup priorities.
MIT — see LICENSE.