About Product Graph
An AI-agent-readable knowledge graph of your product — its features, personas, agent skills, and standards — that agents keep current and humans keep correct.
Why it exists
Coding agents are only as good as the context they're given — and today that context lives in docs and skills.mdfiles that drift the moment they're written. Product Graph replaces that with a single structured graph agents read over MCP, and that they help maintain by proposing changes as they build. A human review queue keeps it accurate, so the knowledge stays trustworthy instead of rotting.
Built for two sides
For coding agents
Agents read your product graph over MCP to get accurate context — features, APIs, reusable skills, and the standards they must follow — and propose updates as they build.
For teams & maintainers
Every agent-proposed change lands in a review queue with full provenance. Leads and PMs approve what goes live, so the graph stays trustworthy instead of rotting.
How it works
Connect via MCP
Point Claude Code, Cursor, or any MCP client at your workspace with a token.
Agents propose
Your agent scans the repo and submits nodes and edges as a structured diff.
Humans approve
Proposals are reviewed and approved; only then do they become part of the live graph.
What makes it different
A graph, not docs
Structured, queryable product knowledge — not stale wikis or scattered skills.md files.
Team-wide, not per-product
A workspace is your team’s graph spanning all its products. Define an agent skill or standard once and every product — and every agent — inherits it via MCP. Cross-product relationships live in one place.
MCP-native
Built for agents to read and write directly, with humans curating the result.
Governed by review
A pending queue + audit trail keeps an auto-generated graph from drifting.
Get in touch
Product Graph is built by VibeLevel.
Get started with Product Graph
Browse public graphs, read the docs, or map your own product.