Feature
Your compiled knowledge, available to every AI that asks for it.
Your compiled knowledge graph — deterministic, source-anchored, contradiction-preserving — available to any AI tool via Model Context Protocol. Instead of pasting raw documents into a prompt, your AI gets the compiled version: the minimum sufficient context, with contradictions flagged and provenance intact.
Why this matters
Making your graph available to AI agents turns structured knowledge into operational context.
MCP integration is in progress. Join the waitlist to be notified when it ships.
What it does
MCP integration makes your compiled knowledge consumable by any AI that supports the protocol.
Capability 01
Connect any MCP-compatible AI to your compiled SILKLEARN workspace
Capability 02
Skills and extensions for Cursor, Claude, Copilot, and Windsurf
Capability 03
Agents get the minimum sufficient context — contradictions flagged, provenance intact — not a raw document dump
In depth
Most AI tools get context wrong. They retrieve document chunks based on query similarity — which means they bring in fragments, miss dependencies, and include contradictions the user doesn't know about. SILKLEARN's MCP integration changes this. Your compiled knowledge graph — deterministic, contradiction-preserving, source-anchored — becomes an MCP server that any AI client can connect to. Claude, Cursor, GitHub Copilot, and Windsurf can pull exactly the nodes required for the task at hand, with contradictions flagged and provenance intact. The agent doesn't guess at context. It gets the compiled version.
How it works
Connect your graph to AI agents via MCP.
- 01
Compile your knowledge first
Run your documents through SILKLEARN. The result is a compiled knowledge graph — every claim source-anchored, every contradiction flagged, every gap named.
- 02
Enable MCP server
With one toggle, your compiled workspace becomes an MCP server. Any AI tool that supports the Model Context Protocol can connect to it.
- 03
Install the skill or extension
Install the SILKLEARN skill for your preferred tool — Cursor, Claude Desktop, GitHub Copilot, or Windsurf. The skill knows how to query your compiled workspace.
- 04
AI pulls compiled context
When your AI agent needs context, it queries SILKLEARN instead of raw documents. It gets the minimum sufficient nodes — compiled, source-anchored, contradiction-preserving — for whatever task it is working on.
- 05
Contradictions are preserved, not erased
SILKLEARN never averages contradictions away. Each side is kept and branched, source-anchored, so your AI sees the disagreement — and where each claim comes from — instead of a silent compromise.
Common questions
- What is MCP?
- Model Context Protocol is an open standard that lets AI clients (like Claude or Cursor) connect to external data sources and tools. When SILKLEARN exposes your compiled knowledge as an MCP server, your AI tools can query it directly — compiled, source-anchored context instead of raw document retrieval.
- Which AI tools will be supported?
- Any tool that supports MCP — Claude Desktop, Cursor, GitHub Copilot (via extensions), and Windsurf at launch. We are also building dedicated skills and plugins for each tool for the best integration experience.
- When is this available?
- MCP integration is coming soon. Join the waitlist to be notified when it ships.
- How is this different from giving an AI my documents directly?
- Raw documents give the AI too much or the wrong context. SILKLEARN's compiled version gives it the exact nodes required for the task — source-anchored, with contradictions preserved and flagged. The AI reasons better because the context is better.
Related capabilities
See where AI-ready context changes how agents reason across your knowledge.
Next step