Installation¶

Get the CRP SDK running locally in under five minutes. This page covers Python requirements, package variants, provider setup, and a one-line verification.

Deployment status

Self-hosted CRP is available today. Managed cloud CRP Gateway and CRP Comply are on the managed-cloud waitlist at gateway.crprotocol.io and comply.crprotocol.io.

Requirements¶

Python 3.10 or later
pip 21.0 or later

Install with pip¶

Full (recommended)Core onlyDevelopment

pip install crprotocol[full]

Includes NLP models (GLiNER, spaCy), security features, and all extraction stages.

pip install crprotocol

Minimal install with core protocol, extraction stages 1-2, and basic security.

git clone https://github.com/AutoCyber-AI/context-relay-protocol.git
cd context-relay-protocol
pip install -e ".[dev]"

Provider dependencies¶

CRP is provider-agnostic. Install the SDK for your preferred LLM:

pip install openai          # For OpenAI / Azure OpenAI
pip install anthropic       # For Anthropic Claude
# Ollama: no pip package needed - just run the Ollama server

Verify installation¶

import crp

print(crp.__version__)  # Should print "4.x"

client = crp.SDKClient()
response = client.complete("Hello, CRP!")
print(response.text)
print(f"Risk: {response.crp.risk}, Grounded: {response.crp.grounded}")

Optional dependencies¶

Package	Purpose
`gliner`	Stage 3 Named Entity Recognition
`spacy`	Stage 5 discourse analysis
`sentence-transformers`	Semantic similarity for fact deduplication
`prompt-injection-detector`	ML-based injection detection