Everything you need to manage organizational knowledge
An Obsidian-style wiki with a force-directed graph, semantic search, and native agent integration.
Semantic Search
Vector-powered search across your entire knowledge graph. Agents find relevant information by meaning, not just keywords.
Graph Explorer
Interactive force-directed graph visualization. See how entities, concepts, and sources connect across your organization.
Multi-Source Ingest
Import knowledge from markdown, URLs, or entire Git repositories. Content is automatically structured and linked.
Agent-Native Access
Agents query and update the KB via MCP tools. They reference docs when writing code, reviewing PRs, or answering questions.
Scoped Permissions
Spaces with fine-grained access control. Personal research stays private, team knowledge is shared, agent access is explicit.
Neo4j-Backed Graph
Enterprise-grade graph database under the hood. Handles thousands of pages with sub-second traversal and full-text search.
Interactive Graph Explorer
Visualize your entire knowledge base as a force-directed graph. Nodes represent pages — sized by connection count, colored by type. Click to preview, hover to highlight connections, search to filter.
Three ways to ingest knowledge
Add content from any source. The system extracts entities, detects links, and builds your graph automatically.
Text & Markdown
Paste markdown directly. Entities are extracted, links detected, and the page is placed in your chosen space.
POST /api/v1/wiki/ingest/text
{
"title": "Auth Architecture",
"content": "# Auth Architecture\n\nWe use...",
"page_type": "concept",
"space_id": "eng-team"
}URL Import
Point to any URL. Content is fetched, structured, and linked to existing pages with source attribution.
POST /api/v1/wiki/ingest/url
{
"url": "https://docs.example.com/api",
"space_id": "eng-team"
}Git Repository
Ingest an entire repo or specific file patterns. Directory structure is preserved as page paths.
POST /api/v1/wiki/ingest/repo
{
"url": "https://github.com/org/repo",
"patterns": ["docs/**/*.md"],
"space_id": "eng-team"
}Native Agent Access via MCP
Agents connect to the Knowledge Base through the Model Context Protocol. They search, read, and contribute to your knowledge graph as part of their normal workflow.
wiki_graph_vector_searchSemantic search across all accessible pageswiki_get_pageRead a specific page by pathwiki_graph_neighborsExplore connected pageswiki_ingest_textCreate or update pages# Agent reviews a PR and queries the KB first
> wiki_graph_vector_search("auth middleware")
Found 3 pages:
- auth-architecture (concept, score: 0.94)
- session-token-storage (entity, score: 0.87)
- auth-middleware-comparison (comparison, score: 0.82)
> wiki_get_page("auth-architecture")
# Auth Architecture
We use JWT tokens stored in httpOnly cookies...
Session tokens must NOT be stored in localStorage
per legal compliance requirement LC-2024-03.
# Agent now has context for the review
> "This PR stores the session token in localStorage.
Per our auth architecture docs (LC-2024-03),
tokens must be in httpOnly cookies only."
How teams use the Knowledge Base
Every task your agents execute gets better when grounded in your actual docs.
Code Review
Agent reads architecture docs and design decisions before reviewing a PR, catching violations of documented conventions.
Onboarding
New team members get agents pre-loaded with team knowledge. Ask questions, get answers grounded in your actual docs.
Customer Support
Support agents search product knowledge to answer customer questions accurately, with source links for verification.
Research & Analysis
Agents ingest and cross-reference external sources, building a knowledge graph that grows with every task.