AI & ML interests

Local-first AI agents, on-device inference (llama.cpp, MLX, vLLM), MCP, RAG, human-in-the-loop agent orchestration

Recent Activity

al2m4n  updated a collection 29 days ago
MLX models for TheYgent · Apple Silicon
al2m4n  updated a collection 29 days ago
MLX models for TheYgent · Apple Silicon
al2m4n  updated a collection 29 days ago
MLX models for TheYgent · Apple Silicon
View all activity

Organization Card

TheYgent

Build agents on a canvas. Run them where you point.

TheYgent is a self-hosted, no-code AI agent builder with built-in local inference — private by default, no telemetry, and it runs fully offline. Compose an agent visually (LLM steps, MCP tools, retrieval, routers, guardrails, human approval) and run it on models on your infrastructure: llama.cpp anywhere, MLX on Apple Silicon, vLLM (experimental) on NVIDIA GPUs, or any OpenAI-compatible API you register. Local and hosted models mix freely in one graph, chosen per node and per call.

Website · Docs · GitHub · Videos · Agent templates · Architecture deep-dive

Watch: Voice in, voice out — a no-code AI agent pipeline

Runs models from the Hub

TheYgent installs models straight from Hugging Face through an in-app catalog — every quantization shows whether it fits your machine's memory before you download it. Installed weights register as local files, and the matching engine starts on the first call and unloads when idle; there is no separate model server to run.

  • Chat and vision — GGUF via llama.cpp, MLX on Apple Silicon
  • Embeddings for retrieval (RAG) into your own pgvector
  • Speech-to-text and text-to-speech for voice agents
  • Image generation

Browse models that run in TheYgent:

What's in the box

  • Visual builder — the graph is the agent; publishing mints an immutable, content-addressed version you can roll back to.
  • Tools & MCP — built-in tools, HTTP tools, and any MCP server over stdio or HTTP.
  • RAG — crawl a site or upload documents, embed locally, wire retrieval into any graph.
  • Durable runs and human-in-the-loop — runs resume from the last completed step after a crash; approval gates can wait for days.
  • Observability — a span waterfall for every run, stored in your own Postgres.
  • Private by default — no telemetry, usage analytics, update checks, or vendor account. Every connection it makes on its own is listed, with when and why.

Quickstart

git clone https://github.com/al2m4n/TheYgent.git
cd TheYgent
cp .env.example .env
make up   # then open http://localhost:5174

Self-hosts on macOS and Linux — bare metal, Docker Compose, or Kubernetes. Fair-code under the Sustainable Use License.

models 0

None public yet

datasets 0

None public yet