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

# Knowledge Bases

> Understand how Knowledge Bases differ from conversational memory and when to use them.

If memory is what your AI learns from talking to you, a Knowledge Base is the shelf of books you hand it on purpose.

## What this means in practice

Knowledge Bases are best for stable reference material such as:

* docs
* policies
* runbooks
* imported project notes
* scoped reference data for a team, agent, or workspace

They are meant to sit **beside** conversational memory, not replace it.

## How it actually works

The main implementation and HTTP surfaces live in:

* `convex/crystal/knowledgeBases.ts`
* `convex/crystal/knowledgeHttp.ts`
* `packages/mcp-server/README.md`
* `convex/schema.ts`

Knowledge Base operations include listing, creating, importing, querying, and background enrichment/backfill flows.

## Commands / examples

Representative tool names:

* `crystal_list_knowledge_bases`
* `crystal_query_knowledge_base`
* `crystal_import_knowledge`

Representative endpoints:

* `GET /api/knowledge-bases`
* `POST /api/knowledge-bases/:knowledgeBaseId/import`
* `POST /api/knowledge-bases/:knowledgeBaseId/query`

## Common mistakes

* treating Knowledge Bases as just another name for memory stores
* mixing mutable conversational memories and stable reference imports together conceptually
* documenting import/query flows without tying them to the actual HTTP/tool surface

## Source of truth

Primary files behind this page:

* `convex/crystal/knowledgeBases.ts`
* `convex/crystal/knowledgeHttp.ts`
* `packages/mcp-server/README.md`
* `convex/schema.ts`