The git for organizational context.
An open protocol for synchronizing organizational knowledge between human and AI actors. Content-addressed. Versioned. Permissioned. Auditable. Real-time.
Why we need this.
AI agents are entering organizational workflows faster than the infrastructure to support them.
Information decay
A policy changed this morning is not reflected in the agent's working memory. Decisions are made on outdated information. In regulated industries, this is a compliance risk.
Coordination collapse
Multiple humans and agents writing to the same knowledge base with no protocol governing ordering, conflict detection, or attribution. An agent overwrites a human's update. A human unknowingly reverts an agent's correction.
Auditability gap
Regulators ask: which version of which document informed this agent's decision? Current systems cannot answer. There is no audit trail that crosses the human-AI boundary.
Five primitives.
MCP connects agents to tools. ContextSync connects agents to organizational context.
Content-addressed artifacts
Stable ctx:// URIs that persist across edits, renames, and moves.
Versioned state
Every mutation is a new version with an author, timestamp, and diff. Rollback is free.
Change subscriptions
SSE-based push notifications. No polling. Millisecond propagation.
Permissioned access
Default-deny tuples for humans and agent classes. Glob patterns for bulk grants.
Provenance tracking
Every read and write captured. The audit trail regulators ask about.
See it run.
Three ways to experience a live protocol server with real data from three industry simulations.
Simulation Theatre
Watch scripted narratives play out: humans and agents reading and writing to the same artifacts in real time. Actor cards pulse as each beat fires. The audit log fills live.
Run a simulation →AI Playground
Chat with a Groq-powered agent that reads organizational context via the protocol. Every tool call is a real API request that produces a real provenance record.
Talk to an agent →Explorer Dashboard
Browse every artifact, org chart, permission grant, and provenance record. Drill into any artifact for its full version timeline and diff viewer.
Open the dashboard →Built with
Reference implementation stack.
Enterprise ready
The same protocol on production infrastructure.
Open protocol. Managed platform.
The protocol is open source and always will be. The commercial opportunity is a fully managed platform that eliminates infrastructure complexity for teams that want to adopt without operating.
Self-hosted protocol
- --Full protocol spec and reference server
- --CLI and dashboard included
- --Single-node SQLite deployment
- --Apache 2.0 license, no restrictions
- --Community support via GitHub
ContextSync Cloud
- --Multi-region, auto-scaling infrastructure
- --SSO / SAML / OAuth 2.0 authentication
- --Compliance packages (SOC 2, HIPAA BAA)
- --Native connectors: SharePoint, Confluence, Notion, Drive
- --Advanced provenance analytics and drift detection
- --Cross-org federation and SLA guarantees
| Tier | Target | Scale | Includes |
|---|---|---|---|
| Starter | Startups, small teams | Up to 20 humans + 10 agents | Hosted server, dashboard, community support |
| Team | Mid-market, departments | Up to 100 humans + 50 agents | SSO, connectors, priority support, 99.9% SLA |
| Enterprise | Large organizations | Unlimited, multi-region | Federation, compliance, dedicated onboarding, custom SLA |
Teach any AI to implement ContextSync.
A single prompt that gives Claude, ChatGPT, or any AI assistant complete knowledge of the protocol. Copy it, paste it, and ask for help building your integration.