Research and experience graphs
How evidence and experience entities relate, short IDs, and why the two graphs stay linked.
Every Crux project maintains two linked graphs. Ask reads both, which is why its answers can connect what you learned to what you are building.
Research graph
Research holds the evidence layer. Its item kinds cover the full arc of product research:
| Kind | Holds |
|---|---|
| Frame, Question, Assumption, Evidence, Gap | The structure of your enquiry |
| Note, Highlights, File | Raw material and what you pulled from it |
| Persona, Journey, Flow, Feature | Synthesis of who, what, and how |
| User test, Table | Structured findings and research tables |
Items appear in the Research tab; open one to read, edit, and ask questions in context. See Research graph for the detail.
Experience graph
Experience holds the product you are building: Page, Feature, Component, Content, Menu, and Sitemap entities. Each has a stable short ID such as ex-156132, a description of why it exists, and evidence links back into research. Those short IDs are what MCP tools and the browser overlay use to connect code and live UI to the why-story. See Experience graph.
First analysis
The last step of project setup seeds both graphs: the Research agent builds a frame with Questions, Assumptions, Evidence, and Gaps, the Experience agent drafts the first entities, and an orchestration layer links the two. Both run in view, with a progress row and an expandable trace each. You keep enriching either side afterwards with Ask, uploads, or MCP writes.
Finding your way around
Use Search (⌘K) to jump to any item in either graph — see Search across your project. For the bigger picture of how the surfaces fit together, see What is Crux?.
Next
Continue with Applying and materialising answers, or see Analyse & Label for turning a single source into highlights and a frame.
Related articles
Documents: First-analysis agents · Research graph · Experience graph