Skip to content

CausalIQ Knowledge User Guide

Overview

The causaliq-knowledge package is part of the CausalIQ ecosystem for intelligent causal discovery and inference. CausalIQ Knowledge provides the following capabilities for integrating LLM knowledge into causal discovery workflows:

Capability Pattern Description
generate-graph Create Generate causal graphs from network context using LLMs

CLI-Only Utilities

The following commands are available only through the CLI and are not accessible as workflow actions:

Command Description
list-models List available LLM models from configured providers
cache-stats View LLM cache statistics and costs
export-cache Export LLM cache entries to files
import-cache Import LLM cache entries from files

Command Line, Workflow or Programmatic Access

As with all CausalIQ packages, users may access the capabilities of CausalIQ Knowledge in three ways:

  • Command Line Interface (CLI) provides an easy introduction to generating graphs from the command line. This is primarily orientated to generating a single graph and thus gaining an initial understanding of how to use the capability.

  • CausalIQ Workflows allows users to include CausalIQ Knowledge steps within workflows which can combine learning graphs from data or LLMs, performing inference, analysing results, through to generating publication tables and charts.

  • Programmatic Access using the Python API for complete flexibility over the processing logic.

The CLI and workflow routes use the same command or action name respectively, which in turn matches the capability name in the table above. Parameters are named identically, and as far as practical, the capability behaviour is the same in the CLI and workflow interfaces.

Workflow Concepts

For common workflow concepts that apply across all CausalIQ packages, see the CausalIQ Workflow User Guide:

The individual action guides in this section document knowledge-specific parameters and behaviour for each capability.

LLM Provider Setup

CausalIQ Knowledge uses direct vendor-specific API clients (not wrapper libraries) to communicate with LLM providers. This approach provides reliability and minimal dependencies. Currently supported:

Provider Environment Variable Free Tier Console URL
Groq GROQ_API_KEY Yes (fast) console.groq.com
Google Gemini GEMINI_API_KEY Yes aistudio.google.com
OpenAI OPENAI_API_KEY No platform.openai.com
Anthropic ANTHROPIC_API_KEY No console.anthropic.com
DeepSeek DEEPSEEK_API_KEY No platform.deepseek.com
Mistral MISTRAL_API_KEY No console.mistral.ai
Ollama (local) Yes ollama.ai

Use cqknow list-models to see which providers are configured and what models are available.

Storing API Keys

Set environment variables for your chosen providers:

# PowerShell
$env:GROQ_API_KEY = "your-api-key"
$env:GEMINI_API_KEY = "your-api-key"

# Bash
export GROQ_API_KEY="your-api-key"
export GEMINI_API_KEY="your-api-key"

Free Options

Groq offers a generous free tier with extremely fast inference:

  1. Sign up at console.groq.com
  2. Create an API key
  3. Set the environment variable
  4. Use in code:
cqknow generate-graph -n context.json -o results/ -c cache.db \
    -m groq/llama-3.1-8b-instant

Google Gemini (Free Tier)

Google offers free access to Gemini models:

  1. Sign up at aistudio.google.com/apikey
  2. Create an API key
  3. Set GEMINI_API_KEY environment variable
  4. Use in code:
cqknow generate-graph -n context.json -o results/ -c cache.db \
    -m gemini/gemini-2.5-flash

Ollama (Local)

Run models locally with Ollama (no API key needed):

  1. Install Ollama from ollama.ai
  2. Pull a model: ollama pull llama3
  3. Use in commands with -m ollama/llama3

What's Next?