Skip to content

Getting Started

Installation

Install Sententia and its dependencies:

pip install sententia

Run with OpenAI

python -m sententia /path/to/markdown/docs \
  --llm-protocol openai \
  --llm-url https://api.openai.com \
  --llm-model gpt-4 \
  --llm-token sk-...

Run with Ollama

Ollama works without an API key:

python -m sententia /path/to/markdown/docs \
  --llm-protocol ollama \
  --llm-url http://localhost:11434 \
  --llm-model llama3

Run with Anthropic

python -m sententia /path/to/markdown/docs \
  --llm-protocol anthropic \
  --llm-url https://api.anthropic.com \
  --llm-model claude-3-sonnet \
  --llm-token sk-ant-...

MCP Server Mode

Add the --mcp flag to start in MCP Server mode instead of REST API:

python -m sententia /path/to/markdown/docs \
  --llm-protocol ollama \
  --llm-url http://localhost:11434 \
  --llm-model llama3 \
  --mcp

Configuration via .env file

Instead of passing all parameters via CLI, you can use a .env file:

# .env
SENTENTIA_LLM_PROTOCOL=ollama
SENTENTIA_LLM_URL=http://localhost:11434
SENTENTIA_LLM_MODEL=llama3

python -m sententia /path/to/markdown/docs

Use --env-file to specify a custom path:

python -m sententia /path/to/markdown/docs --env-file /path/to/config.env

See Configuration for all available environment variables and priority rules.

First Request

After starting in REST mode:

curl -X POST http://localhost:8000/search \
  -H "Content-Type: application/json" \
  -d '{"query": "how to set up authentication", "top": 5}'

Ask (RAG)

curl -X POST http://localhost:8000/ask \
  -H "Content-Type: application/json" \
  -d '{"query": "how to set up authentication?"}'

Files

curl http://localhost:8000/files/docs/auth.md