> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gdilabs.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Grounded in your knowledge

> Every Cerebrum answer is anchored to your organization's documents, not the model's generic training data. Cite-and-source is built in.

## The knowledge hub

A library of organizational documents — product notes, project briefs, partner threads, PDFs, READMEs, hand-written guidance, and decisions agents have made along the way. Each document carries metadata:

* **Scope** — where in the organization it lives (a product, a project, a team).
* **Sensitivity** — public, partner, internal, or secret.
* **Source** — where the content originally came from.
* **Owner and team** — who keeps it current.

## How agents use it

When an agent answers, the platform pulls the most relevant chunks from the hub — filtered by sensitivity tier, scope, and recency — and includes them in the prompt. The agent answers from those chunks, with citations back to the source documents.

If you ask "what's our deployment story for Project X?", you'll get an answer pulled from the actual Project X documents, with references — not a guess based on what "deployment" usually looks like.

## Working with the knowledge hub

* **From Claude Desktop or Claude Code.** Install the [Knowledge Hub MCP server](/developers/components/mcp-cerebrum-kb). It exposes four tools: search, fetch, list, and node-with-neighbors.
* **From the dashboard.** The admin surface lets you edit documents directly. Edits sync to the search index within about a minute.
* **From agents themselves.** Workers record their decisions as they happen, so today's reasoning becomes tomorrow's searchable context.