Structured Memory for
Human-AI Teams
Knowledge Objects (KO) turn scattered decisions, insights, and patterns into a structured, verifiable knowledge base. Built for teams where humans and AI agents collaborate on complex projects.
Quick Start
npm install -g ko-viewer-cliko loginko initInstalls the KO plugin for Claude Code. Restart Claude Code after setup to activate KO skills.
AI agents lose context between sessions. Team knowledge lives in chat logs nobody reads. Decisions get re-debated because nobody recorded the reasoning. KO fixes this.
Structured Knowledge
Every Knowledge Object follows a five-attribute schema: type, status, confidence, source, and scope. No more loose notes — every piece of knowledge is classified, traceable, and verifiable.
Quality Control
Four-state lifecycle (review → verified / disputed / archived) ensures knowledge is validated before it's trusted. Track who verified what, when, and why.
AI-Native Integration
Claude Code plugin auto-loads KOs into agent context. Agents can recall, create, and verify knowledge objects — bridging human insight and AI capability.
Full Traceability
Every KO records its origin: which channel, which agent, which conversation. Usage statistics track retrieval counts, helpfulness, and feedback over time.
Five Attributes, One Schema
Every Knowledge Object is classified along five dimensions
Type
fact / decision / pattern / insight / skill
Status
review / verified / disputed / archived
Confidence
high / medium / low
Source
channel / agent / conversation
Scope
projects / domains / validity
Access Anywhere
Claude Code Plugin
v0.3Auto-loads KOs into agent context. Use /ko-recall to search, /ko-store to create, /ko-verify to validate. 3 hooks + 4 skills.
KO Space
Web AppBrowse, search, and filter Knowledge Objects with a visual interface. Full-text search, type/status/project filters, and backlink navigation.
Enter KO Space →