How the CoreWeave Documentation Is Organized, and Where to Start

Start with the Documentation Index on the Forge Home page. It splits the docs into two top-level areas — Infrastructure and Forge — and each area is broken into task-based sections. If you are new, read the Infrastructure Platform section first to get set up; if you already have a cluster and want to build or evaluate models, jump straight to Forge.

The two top-level areas

Infrastructure sections

The Infrastructure side is organized by the layer you are working at:

• Platform — Get started, then choose instances and capacity, regions and Availability Zones, billing, and policies. This is the natural first stop for a new account.
• Mission Control — The Mission Control Agent and the Mission Control MCP server, used for investigating your environment and making supported changes.
• Infra Control — CoreWeave Kubernetes Service (CKS) plus the storage and networking products your clusters use.
• Observability — Grafana dashboards, logs, metrics, telemetry, and alerts for monitoring fleet hardware and workloads.
• Workloads — SUNK for Slurm on Kubernetes, CoreWeave Inference for serving models, and CoreWeave Sandbox for isolated environments.
• Access Control — Accounts and authentication, IAM access policies, automated user provisioning, and trust and compliance.

Forge sections

Forge is the development and evaluation side, grouped by stage of the model or agent lifecycle:

• Weights & Biases — W&B Models for experiment tracking, hyperparameter sweeps, artifacts, reports, and automations, with W&B Weave alongside for LLM and agent evaluation.
• ARIA — An AI research agent that analyzes experiments, explains results, and builds visualizations and reports. Marked as Preview.
• Agent Lens — Intelligent agent observability and improvement.
• Notebooks — Reactive Python notebooks for exploring data, prototyping model workflows, and analyzing runs and artifacts.
• Registry — A versioned hub for models and datasets with lineage and deployment handoff.
• Serverless services — Serve, fine-tune, and run models on CoreWeave infrastructure without managing it. This includes Post-Training (Serverless RL and SFT, plus distilling specialized models from real application traffic), Serverless Inference (open-source foundation models through an OpenAI-compatible API, integrated with Weave), and Serverless Sandbox (on-demand, isolated compute environments you create and discard from Python).

Where to start, by task

• Brand-new account, nothing deployed yet → Infrastructure → Platform. Work through get started, then instances and capacity, then regions and Availability Zones.
• Need a Kubernetes cluster → Infrastructure → Infra Control (CKS), then the storage and networking pages for the products your cluster uses.
• Something looks wrong in your environment → Infrastructure → Mission Control, and Observability for dashboards, logs, metrics, and alerts.
• Running Slurm or serving a model → Infrastructure → Workloads.
• Managing users and permissions → Infrastructure → Access Control.
• Tracking experiments or evaluating an LLM/agent → Forge → Weights & Biases (Models and Weave).
• Want an agent to analyze your experiments → Forge → ARIA (Preview).
• Prototyping in a notebook → Forge → Notebooks.
• Serving or fine-tuning without managing infrastructure → Forge → Serverless services.

Practical notes

• The index page itself is the fastest map: each section is a short labeled block, so you can scan headings rather than open pages one by one.
• ARIA is labeled Preview, so treat its behavior as subject to change.
• The page shows a last-modified date of September 29, 2026, which is a useful signal for how current a given section is — but check individual pages, since they can be updated independently of the index.