NativeLab light mode workspace

Feature catalogue - v0.3.7

Everything in one local workspace.

NativeLab covers the full loop: install a runtime, register models, chat, attach documents, build pipelines, run Labs, expose endpoints, and script the same engine from a terminal.

Latest

Recent work focused on pipelines, setup, and stability.

v0.3.7 adds AI-assisted pipeline generation, native pipeline helper paths, sidebar resizing, centralized LLM error dialogs, and refreshed first-run setup flows.

AI Pipeline Builder

Generate validated pipeline JSON from a plain-English request, then load it for testing.

new

Native helpers

C/Rust acceleration for deterministic graph, validation, prompt, sampler, and detection hot paths.

optional

Auto setup

Hardware-aware, resumable setup with llama.cpp or HF Transformers backend choices.

resumable

UI hardening

MainWindow split, centralized QThread cleanup, context meter, and plain-language LLM errors.

safer

App surfaces

Use the same engine from GUI, CLI, Labs, and integrations.

Every surface routes through shared runtime state so model loading, context, API profiles, and saved pipelines stay consistent.

NativeLab dev workspace

Dev workspace

Labs, logs, integrations, pipelines, MCP, and skills live behind Developer Mode.

NativeLab skills tab

Skills and Labs

Shared skill injection plus drop-in Labs panels using one endpoint API.

NativeLab server controls

Runtime controls

llama.cpp paths, context, GPU offload, ports, and backend-specific settings.

Core capabilities

A concise map of what NativeLab does.

The full technical details live in the project docs, but these are the surfaces most users touch first.

Model runtime

  • Local GGUF models
  • llama-server and llama-cli
  • Ollama refs and pulls
  • HF Transformers snapshots
  • OpenAI-compatible APIs
  • Anthropic API profiles

Documents and code

  • Session references
  • PDF summarization
  • Multi-PDF synthesis
  • Source parser for many languages
  • Py-to-Doc project mode
  • Checkpoint resume

Workflow tooling

  • Visual pipeline canvas
  • AI Pipeline Builder
  • Labs endpoint API
  • MCP management
  • Integration HTTP endpoint
  • Discord and WhatsApp profiles

CLI

A terminal control center, not a separate app.

The CLI can chat, load models, run saved pipelines, inspect endpoints, manage skills, run Labs helpers, and serve integration routes.

$ nativelab --cli
Chat / Models / API Models / Skills / Labs / Pipelines / Integrations / Status / Setup

$ nativelab --cli pipeline run research-synthesis --file prompt.txt
$ nativelab --cli endpoint /snapshot --json