Skip to main content

​
Overview

The LLM API provides endpoints for managing Large Language Model integration. LLM features power ontology generation from natural language and entity extraction from unstructured data.

​
List Available Models

Get available LLM models from the configured provider. 
curl -X GET http://localhost:8080/api/llm/models
Response (LLM enabled): 
{
  "provider": "openrouter",
  "enabled": true,
  "models": [
    {
      "id": "anthropic/claude-3-opus",
      "name": "Claude 3 Opus",
      "context_length": 200000
    },
    {
      "id": "openai/gpt-4",
      "name": "GPT-4",
      "context_length": 8192
    }
  ]
}
Response (LLM not configured): 
{
  "provider": "none",
  "enabled": false,
  "models": []
}
​
provider
string
Provider name (openrouter, openai_compat, or none)
​
enabled
boolean
Whether LLM features are available
​
models
array
Available model list (cached with TTL)

​
Configuration

LLM integration is configured via environment variables:
VariableDescriptionExample
LLM_ENABLEDEnable LLM featurestrue
LLM_PROVIDERProvider typeopenrouter or openai_compat
LLM_API_KEYAPI keysk-...
LLM_BASE_URLBase URL (openai_compat only)https://api.together.xyz/v1
LLM_MODELDefault modelanthropic/claude-3-opus

​
OpenRouter Provider

LLM_ENABLED=true
LLM_PROVIDER=openrouter
LLM_API_KEY=sk-or-v1-...
LLM_MODEL=anthropic/claude-3-opus

​
OpenAI-Compatible Provider

LLM_ENABLED=true
LLM_PROVIDER=openai_compat
LLM_API_KEY=your-api-key
LLM_BASE_URL=https://api.together.xyz/v1
LLM_MODEL=meta-llama/Llama-3-70b-chat-hf

​
External LLM Providers

Dynamically install custom LLM provider plugins from Go source repositories.

​
List External Providers

Get all installed external LLM providers. 
curl -X GET http://localhost:8080/api/llm/providers
Response: 
[
  {
    "name": "ollama",
    "description": "Local Ollama instance",
    "repository": "https://github.com/example/mimir-llm-ollama",
    "version": "v1.0.0",
    "installed_at": "2026-03-01T10:00:00Z",
    "loaded": true
  }
]
​
name
string
Provider identifier
​
loaded
boolean
Whether the provider .so is currently loaded

​
Install External Provider

Install an external LLM provider from a Git repository. 
curl -X POST http://localhost:8080/api/llm/providers \
  -H "Content-Type: application/json" \
  -d '{
    "name": "ollama",
    "repository": "https://github.com/example/mimir-llm-ollama",
    "version": "v1.0.0",
    "description": "Local Ollama instance"
  }'
​
name
string
required
Unique provider identifier
​
repository
string
required
Git repository URL. Must export var Provider satisfying the LLM provider interface.
​
version
string
required
Git tag, branch, or commit SHA
​
description
string
Human-readable description
Response: 201 Created or 422 Unprocessable Entity (compilation failure)

​
Get External Provider

Get details for a specific external provider. 
curl -X GET http://localhost:8080/api/llm/providers/ollama

​
Uninstall External Provider

Remove an external LLM provider. 
curl -X DELETE http://localhost:8080/api/llm/providers/ollama
Response: 204 No Content
Go plugins cannot be unloaded from memory. An orchestrator restart is required for full removal.

​
LLM Features

LLM integration powers:
  • Ontology Generation: Generate OWL ontologies from natural language descriptions
  • Entity Extraction: Extract structured entities from unstructured text, documents, and media
  • Schema Inference: Infer entity types and relationships from raw data
See:

Build docs developers (and LLMs) love

Get started for freeTalk to us