KNOWLEDGE + LEARNING INFRASTRUCTURE

KNOWLEDGE FABRIC

A provenance-aware knowledge graph for autonomous software. Live capabilities and registered evidence stay on the truth plane. Privacy-safe routing history and verified execution outcomes live on a separate learning plane that can improve ranking without becoming fact by repetition.

Truth plane

Current SqueezeOS capabilities, routes, payment constraints and explicitly registered evidence remain the factual authority.

Learning plane

Aggregate routing observations and verified execution outcomes can add ranking confidence and explainability without rewriting facts.

Bounded influence

Learning is a tie-breaker only. It cannot override materially stronger relevance or hard payment and budget constraints.

Truth / learning separation

Learning cannot promote a claim to fact. A successful payment settlement is not automatically execution proof. Accepted delivery plus successful settlement can support verified execution; a later settlement failure does not by itself prove the handler execution failed. Public surfaces expose aggregate evidence only, not raw ECHO demand, agent fingerprints, prompts or secrets.

Live Knowledge Fabric

This panel reads the production Fabric through a same-origin proxy. It never invents fallback graph data.

Loading live status…
Enter an intent to query the live catalog.

Machine interface

STATUS GET https://squeezeos-api.onrender.com/api/knowledge-fabric/status CATALOG GET https://squeezeos-api.onrender.com/api/knowledge-fabric/catalog?q=<intent> GRAPH GET https://squeezeos-api.onrender.com/api/knowledge-fabric/graph?q=<intent> EXPLAIN GET https://squeezeos-api.onrender.com/api/knowledge-fabric/explain?capability=<capability> MCP POST https://squeezeos-api.onrender.com/mcp tool: knowledge_fabric_query AGENTROUTES GET|POST https://squeezeos-api.onrender.com/api/agent-routes/discover