yeeeeezus · production portfolio

Agent Fleet

Live AI infrastructure on Hugging Face Spaces — real models, real inference, no mocks — plus a one-command Docker fleet of the agent stack (n8n, Flowise, Langflow, Open WebUI, AnythingLLM, LobeChat, OpenMuse) ready to launch the moment Docker hosting is attached.

status 4 live · 7 staged account huggingface.co/yeeeeezus source github.com/yeeeeezus/agent-fleet last verified 2026-10-02

Live now — free, no keys, no signup

Real inference that runs in your browser. Weights stream once from the HF CDN and are cached locally.

💬Chat — Local AILIVENEW

Production AI chat, 100% client-side: llama.cpp via WebAssembly in a Web Worker, streaming tokens, stop/regenerate, conversations, system prompts, real tok/s stats, Markdown + code rendering, PWA.

wllama · Qwen3 1.7B/0.6B · Qwen2.5 1.5B/0.5B · Llama 3.2 1B · Q4_K_M

⚔️Edge ArenaLIVE

Blind A/B evaluation for sub-1B open models: two anonymous fighters answer side-by-side, you vote, live Elo. Entirely in-browser — one Web Worker per fighter.

wllama · Supra2-100M · Qwen3-0.6B · session Elo

🧪AI LabLIVE

Transformers.js playground: streamed text generation with chat templates, sentiment analysis, and named-entity recognition — WebGPU when available, WASM otherwise.

transformers.js · SmolLM2-135M · Qwen2.5-0.5B · DistilBERT · BERT-NER

⚡Inference IndexLIVE

Semantic search over the 21-tool aisearch index (verified against live GitHub / HF APIs) plus the servable model catalog for this account. Query embeddings run in-tab.

all-MiniLM-L6-v2 · WASM · verified 2026-09-30

Staged — full agent stack, one command away

Fully packaged Docker Spaces: Dockerfiles, configs, secrets and an idempotent provisioner.

🔁n8nPRO-READY

Workflow automation with 400+ integrations and AI agent nodes. Webhook URL pre-wired, encryption key managed as a Space secret, owner account bootstrapped on deploy.

docker · port 7860 · secrets: N8N_ENCRYPTION_KEY

🤖Local Chat — Open WebUI + OllamaPRO-READY

Zero-config chat with baked-in models: llama3.2:1b, qwen2.5:0.5b, nomic-embed-text for RAG. Includes an OpenAI-compatible API endpoint for the rest of the fleet.

docker · supervisord · models baked at build

🧠OpenMusePRO-READY

CopilotKit/OpenMuse — the personal agent with browser, terminal, files and durable work. Full stack: nginx + API (PGlite) + task worker + Playwright Chromium browser worker. Needs a free CopilotKit project key.

docker · node 24 · pnpm 11.19 · playwright

🧩FlowisePRO-READY

Visual LLM agent builder: drag-and-drop agents, tools, memory, RAG pipelines; every chatflow becomes a REST API. Basic-auth gated via Space secrets.

docker · node 20 · secrets: FLOWISE_USER/PASS

🌊LangflowPRO-READY

Visual agent & workflow builder; every flow is an instant REST endpoint. Superuser provisioned from a Space secret on first boot.

docker · port 7860 · secret: LANGFLOW_SUPERUSER_PASSWORD

📚AnythingLLMPRO-READY

All-in-one RAG workspace: drop documents into workspaces, chat with citations, built-in AI agents (web-browsing, SQL, custom).

docker · port 3001 · first-run wizard

💠LobeChatPRO-READY

Polished ChatGPT-style UI with agent marketplace, plugins, vision and artifacts. Access-code gated; API keys stay client-side.

docker · port 3210 · secret: ACCESS_CODE

Why staged? Hugging Face now requires a PRO subscription to host Docker/Gradio Spaces on free hardware (verified: 402 Payment Required on api/repos/create, 2026-10-02). Everything is packaged and idempotent — attach PRO (or any Docker host) and run python deploy.py from the repo to bring all seven up in ~15 minutes. Static Spaces, like the four live ones above, remain free.

The top-20 agent landscape

Where each popular platform lives in this fleet.

Platform	What it is	Status in this fleet
n8n	Workflow automation + AI agents	staged · docker
OpenMuse	Personal agent (browser/terminal/files)	staged · docker
Flowise	Visual agent builder	staged · docker
Langflow	Visual agent builder	staged · docker
Open WebUI	Self-hosted chat UI + Ollama	staged · local-chat
AnythingLLM	RAG workspace + agents	staged · docker
LobeChat	Chat platform + agent market	staged · docker
In-browser LLMs	wllama / transformers.js	live · 4 spaces
Dify	LLM app platform	multi-service · needs compose host
OpenHands	Autonomous dev agent	needs Docker-in-Docker
AutoGPT	Autonomous agent platform	needs DB+Redis+Meili
CrewAI / AutoGen	Agent orchestration frameworks	libraries — used inside n8n/Flowise
Letta	Memory agents (MemGPT)	v1 server archived; local-first now
LibreChat	Multi-provider chat UI	needs MongoDB