Ollama

Ollama Plugin

Stands up a local Ollama container and routes nSelf's AI features through it at zero API cost. Free — MIT licensed.

Install

nself plugin install ollama

No license key required.

Description

One-click offline LLM stack. Stands up an Ollama Docker container, auto-pulls gemma-3-4b on first start, and registers as a provider in plugin-ai. All nSelf AI features route through Ollama when NSELF_AI_PROVIDER=ollama. Zero cloud dependency, zero API key, zero usage cost after install. GPU passthrough is optional and localhost-only (port 11434 binds to 127.0.0.1 and is never exposed externally).

This plugin runs as its own container in your nSelf stack (rebuild with nself build && nself start after install).

Category: integrations. Current version: 1.1.1.

Configuration

Env Var	Default	Description
NSELF_AI_PROVIDER	(see plugin.json)	Optional.
NSELF_OLLAMA_HOST	(see plugin.json)	Optional.
NSELF_OLLAMA_DEFAULT_MODEL	(see plugin.json)	Optional.
NSELF_OLLAMA_AUTO_PULL	(see plugin.json)	Optional.
NSELF_OLLAMA_GPU	(see plugin.json)	Optional.
NSELF_OLLAMA_CONTEXT_WINDOW	(see plugin.json)	Optional.
NSELF_OLLAMA_TIMEOUT_SECONDS	(see plugin.json)	Optional.
OLLAMA_ENABLED	(see plugin.json)	Optional.
PLUGIN_AI_OLLAMA_MODEL	(see plugin.json)	Optional.
PLUGIN_AI_OLLAMA_URL	(see plugin.json)	Optional.

Ports

Port	Purpose
11434	Ollama service

Database Schema

1 table(s) added to your Postgres database (prefix: np_ollama_):

• np_ollama_model_registry

REST API

GET  /api/tags            — List pulled models (proxied Ollama API)
POST /api/pull            — Pull a model
POST /api/generate        — Generate a completion (proxied Ollama API)

Nginx Routes

Route	Target
(localhost only, no nginx route: 127.0.0.1:11434)	Ollama never exposed externally

Examples

Check health

curl http://localhost:11434/health

Source

plugins/ollama/

Manifest: plugins/ollama/plugin.json

See Also

• Model — manage local Ollama models
• AI-CLI — chat + model management

← Home →