Local AI Packaged
github.com/coleam00/local-ai-packaged- Category
- AI Tools
- Rank
- No. 975Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- coleam00
- GitHub
- 3.8k stars
- Date
About
Cole Medin's batteries-included local AI bundle — Ollama, Supabase, n8n, Open WebUI and more in one self-hostable Docker stack.
What it does
It gives builders a private AI workflow workbench on their own Docker host. Local models feed chat and low-code agent tools, while database, authentication, retrieval, graph storage, search, observability, and HTTPS services share one environment. Included workflows provide starting points for retrieval-based agents.
Why it's ranked here
The package covers an unusually broad path from model inference to agent building, data, search, tracing, and browser chat. Its practical value comes from working integrations and starter workflows, not merely a service list. However, mutable container tags and runtime dependency cloning make repeatable operations harder than the convenient setup suggests.
What's good
Hardware profiles cover CPU, Nvidia, and AMD configurations, with a separate path for running the model service directly on a Mac. Private deployment binds service ports to localhost, while public deployment removes direct Supabase port exposure. Starter retrieval workflows, persistent volumes, automatic search-secret generation, and an authenticated chat-to-workflow bridge shorten experimentation.
Tradeoffs
Operators must create and protect numerous credentials before production use. Several containers track latest or main images, and setup updates the Supabase master branch at runtime, so identical commands may produce different stacks later. Startup waits a fixed ten seconds instead of checking Supabase readiness. Mac Docker deployments cannot expose Apple Silicon GPUs to the model container.
How to use it well
Use it for local prototypes and self-hosted experiments that combine chat, agent workflows, retrieval, databases, search, and tracing. Pick the hardware profile explicitly, replace every example credential, pin external images and dependencies when reproducibility matters, and expose public services only through the proxy. It does not replace managed hosting or remove ongoing security and operations work.
Technical notes+
start_services.py sparse-clones or pulls Supabase master, copies the root environment into supabase/docker/.env, generates the SearXNG key, may rewrite docker-compose.yml, tears down the unified localai project, starts Supabase, sleeps ten seconds, then starts the remaining services. docker-compose.yml uses mutable latest and main tags for several images and preloads Qwen plus Nomic embeddings. docker-compose.override.private.yml binds service ports to loopback; docker-compose.override.public.yml and docker-compose.override.public.supabase.yml reset selected Supabase ports. n8n_pipe.py posts chat input to a bearer-protected webhook without a timeout; its async path makes a synchronous HTTP request, and the no-message branch can reach a return of an unassigned n8n_response.
Observed
- License
- Apache License 2.0
- Packaging
- Docker Compose stack launched by a Python setup script
- Prerequisites
- Python, Git, and Docker or Docker Desktop
- Interfaces
- Browser interfaces, service APIs, n8n webhooks, and an Open WebUI pipe
- Platform profiles
- CPU, Nvidia GPU, AMD GPU on Linux, and no-container Ollama mode for Mac
- Deployment modes
- Private mode exposes loopback ports; public mode limits direct exposure and routes through Caddy
- Dependency structure
- The launcher sparse-clones the Supabase Docker configuration from its master branch at runtime
Read from README.md, n8n_pipe.py, start_services.py, LICENSE, CLAUDE.md, Caddyfile, .env.example, docker-compose.yml, docker-compose.override.public.yml, docker-compose.override.private.yml, docker-compose.override.public.supabase.yml, searxng/settings-base.yml, flowise/create_google_doc-CustomTool.json, n8n-tool-workflows/Create_Google_Doc.json.
What it can do
Run large language models locally
Text prompts and queries → AI-generated text responses
Store and manage AI conversation data
Chat messages and user interactions → Persistent conversation history
Create automated AI workflows
Workflow configurations and triggers → Automated task execution results
Provide web-based chat interface
User messages through web browser → Real-time AI chat responses
Deploy complete AI stack with single command
Docker compose configuration → Running local AI infrastructure
Process data through AI pipelines
Raw data and processing instructions → AI-processed and analyzed data
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.