Where to Find Onyx Documentation: Official Guides and Source References

The primary source for up-to-date Onyx documentation is the official docs site at https://docs.onyx.app, prominently linked in the repository's README.md and continuously generated from the source code.

The onyx-dot-app/onyx repository contains comprehensive documentation both in the form of an online documentation site and inline code references. Whether you are deploying the platform locally or extending its RAG capabilities, the Onyx documentation provides authoritative guidance that reflects the exact version of the code you are running.

Official Documentation Site

The definitive entry point for Onyx documentation is the live site at https://docs.onyx.app. According to the source repository, this link appears in the top-level README.md within the badge section (lines 13-16) and again in the Features section (lines 60-61), ensuring users can always locate the official hub.

The site organizes content around the platform's modular architecture, providing detailed coverage of:

• Core Chat and Agents: Documentation for the FastAPI backend, Celery workers, and React UI collaboration that powers real-time conversational experiences.
• Connectors and RAG: Configuration guides for over 40 data sources, document-fetching pipelines, and hybrid search indexing.
• Deployment Patterns: Step-by-step instructions for Docker Compose, Kubernetes manifests, and cloud-specific Terraform modules (mirroring the deployment/ directory structure).
• Extensibility: Instructions for adding custom Python code, new LLM providers, and third-party action integrations.
• Enterprise Features: Multi-tenant isolation details, role-based access control (RBAC), and SSO implementations found under backend/onyx/auth/.

Navigating Repository Documentation

While the online docs provide user-facing guides, the repository itself serves as the canonical reference for implementation details. Key documentation locations include:

• README.md: The central entry point containing high-level overviews and the official docs link.
• deployment/docker_compose/: Docker Compose configurations for local quick-start deployments.
• backend/onyx/chat/: Core chat engine implementation, agent definitions (backend/onyx/chat/agent.py), and streaming endpoints.
• backend/onyx/connectors/: Implementation details for the 40+ knowledge source connectors.
• background/celery/: Celery app definition and task queue configurations (docfetching, docprocessing).
• cli/: Command-line interface source code for platform management operations.

Practical Implementation Examples

The documentation includes runnable examples for common platform interactions. These snippets assume operation from the repository root.

Local Docker Compose Deployment

For immediate local testing, the documentation references the Docker Compose files in deployment/docker_compose/:

# From the repository root

docker compose up -d

# Once containers are healthy, open the UI

open http://localhost:3000   # macOS; use your browser on other OSes

This launches the full stack including the FastAPI backend, React frontend, and supporting services.

Running Background Workers

Document processing relies on Celery workers defined in background/celery/tasks/docfetching/:

# Activate the Python virtual environment

source .venv/bin/activate

# Start the docfetching worker

celery -A background.celery_app worker -Q docfetching -c ${DOCFETCHING_WORKER_CONCURRENCY:-4}

The ${DOCFETCHING_WORKER_CONCURRENCY:-4} parameter defaults to 4 concurrent workers unless overridden by environment variables.

Creating Custom Agents

The Python SDK allows agent creation as documented in backend/onyx/chat/agent.py:

from onyx.chat.agent import Agent

my_agent = Agent(
    name="Researcher",
    system_prompt="You are a helpful research assistant that can browse the web.",
    tools=["web_search", "code_interpreter"],
)
response = my_agent.run("Summarize the latest AI safety research papers.")
print(response)

The tool registry referenced in the tools parameter lives under backend/onyx/tools/.

Managing Connectors via CLI

For programmatic connector configuration, the CLI source in cli/ supports commands like:

# Register a Slack connector

onyx-cli connector add \
    --type slack \
    --name "Team Slack" \
    --config '{"token":"xoxb-...","channel":"general"}'

Connector handling implementations reside in backend/onyx/connectors/.

Summary

• The official Onyx documentation lives at https://docs.onyx.app, continuously generated from the repository source.
• The README.md in the repository root serves as the primary navigation hub, linking to the docs site and high-level architecture overviews.
• Documentation is organized by architectural layer: Core Chat, Connectors, Deployments, Extensibility, and Enterprise Features.
• Implementation details can be verified against specific directories: backend/onyx/chat/, backend/onyx/connectors/, background/celery/, and deployment/docker_compose/.
• Runnable code examples in the docs cover Docker Compose startup, Celery worker configuration, Python SDK agent creation, and CLI connector management.

Frequently Asked Questions

Where is the Onyx documentation hosted?

The official Onyx documentation is hosted at https://docs.onyx.app. This URL is prominently displayed in the repository's README.md file (lines 13-16 and 60-61) and serves as the primary hub for all platform guides, API references, and deployment instructions.

Does the Onyx documentation match the repository version?

Yes. The documentation site is continuously generated from the repository source code, ensuring that the guides reflect the exact version of the code you are working with. This synchronization applies to all sections including deployment configurations in deployment/ and implementation details in backend/onyx/.

What documentation exists for extending Onyx with custom connectors?

Custom connector documentation covers the backend/onyx/connectors/ directory, which houses implementations for over 40 data sources. The docs provide patterns for adding new knowledge sources, configuring the document-fetching pipeline, and integrating with the hybrid search index. For programmatic management, the CLI documentation in cli/ details how to register connectors using commands like onyx-cli connector add.

How do I find implementation details for Onyx background processing?

Background processing documentation focuses on the Celery implementation under background/celery/. This includes task queue configurations for docfetching and docprocessing workers, concurrency settings, and the interaction between FastAPI backends and asynchronous job processing. Specific worker startup commands are documented with environment variable configurations for production deployments.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}