Skip to main content

Documentation

Knowledge Base

A Neo4j-backed knowledge graph with semantic search, MCP integration, and multi-source ingestion. Build your org's brain and let your agents get smarter with every task.

1

What is the Knowledge Base?

The Knowledge Base stores your organization's documents, decisions, and reference material as interconnected pages in a Neo4j graph database. Every page is vector-indexed for semantic search and linked to related pages through typed relationships. Agents access it natively via MCP — they search before coding, reference docs during reviews, and contribute knowledge after completing tasks.

Semantic Search

Vector-powered search finds pages by meaning. Ask a question in natural language and get the most relevant results.

Graph Explorer

Interactive force-directed visualization shows how pages connect. Nodes are sized by connections and colored by type.

Multi-Source Ingest

Import from markdown, URLs, or entire Git repositories. Content is automatically structured, linked, and indexed.

Agent-Native Access

Agents query and update the KB via MCP tools as part of their normal workflows — code review, research, and more.

Spaces & Permissions

Organize pages into spaces with scoped access. Personal research stays private, team knowledge is shared.

Git Repo Ingest

Point to a GitHub, GitLab, or Bitbucket repo. Markdown files are ingested with directory structure preserved.

Page Types

Entity— Components, APIs, servicesConcept— Patterns, conventions, decisionsComparison— Framework evaluations, trade-offsSource— Imported URLs, repos, references
2

How to Use the KB

Agents connect to the Knowledge Base through the Model Context Protocol (MCP). Add the MCP server URL to your agent's configuration and it gains access to search, read, and write tools.

MCP Server URL

https://wiki.agent.ceo/mcp

Add this URL to your agent's MCP server configuration (e.g. settings.json or claude_desktop_config.json).

Available MCP Tools

wiki_graph_vector_search

Semantic search across all pages in accessible spaces

wiki_graph_vector_search("auth middleware patterns")
wiki_get_page

Read a specific page by its path

wiki_get_page("auth-architecture")
wiki_graph_neighbors

Explore pages connected to a given page

wiki_graph_neighbors("auth-architecture")
wiki_ingest_text

Create or update a wiki page with markdown content

wiki_ingest_text("API Design Guide", "# API Design…", "concept")
wiki_ingest_url

Import content from a URL into the knowledge base

wiki_ingest_url("https://docs.example.com/api")

REST API

The KB also exposes a REST API for programmatic access.

GET/api/v1/wiki/spacesList spaces
POST/api/v1/wiki/spacesCreate space
GET/api/v1/wiki/pagesList pages
POST/api/v1/wiki/ingest/textIngest text
POST/api/v1/wiki/ingest/urlIngest URL
POST/api/v1/wiki/ingest/repoIngest repo
GET/api/v1/wiki/searchSearch pages
GET/api/v1/wiki/graphGraph data
3

Getting Started

Set up your knowledge base in three steps.

1

Create a space

Sign in and go to the KB page. Click "Create a space" — choose personal for private research or org-wide for shared team knowledge.

Get started free
2

Add content

Ingest knowledge via text, URL, or Git repo. The system extracts entities, detects relationships, and builds your graph automatically.

3

Connect an agent

Add the MCP server URL to your agent's configuration. Agents can then search, read, and contribute to the knowledge graph during their workflows.

settings.json
{
  "mcpServers": {
    "wiki": {
      "url": "https://wiki.agent.ceo/mcp"
    }
  }
}

Ready to build your knowledge graph?

Create a space, add your docs, and let your agents get smarter with every task.