Zindex

Diagrams from AI are usually one-offs: redrawn from scratch in the next conversation, with nothing a teammate's Claude or another agent can build on. Zindex keeps them, so your team and your agents keep editing the same diagram. Ask Claude for an architecture, an entity-relationship model, a sequence, a BPMN workflow, a network, an org chart, a UI flow or a flowchart, and Claude saves it in Zindex.

Every change is a new numbered version, and nothing in a diagram's history can be rewritten: each version's hash covers the one before it. Claude can open any version by number or date, see who changed what, and compare any two versions: what was added, changed and removed. Each version's image is kept, so an engine update never changes an old picture.

In a later chat, Claude finds the diagram, reads a short brief instead of the whole drawing, and sends only what changes, so nothing is redrawn by accident. In a team workspace (Team plan), everyone's Claude works on the same diagrams, and so do Claude Code and the agents you build, through the API. History names who saved each version, and an edit made from an older version is refused, so nobody's work is overwritten. Claude can also start from a Mermaid flowchart or erDiagram, and edit an architecture or database diagram as Mermaid text.

Share a diagram when you choose: publish it, or make a share link that shows its images to anyone with it (Pro and Team). The images go into a README, an issue, a doc or a Claude artifact.

What it needs: a Zindex account. Claude asks you to sign in the first time a tool needs one; checking or previewing a diagram works without signing in. Free to start; $5 a month for unlimited diagrams.

What it doesn't do: Zindex runs no model on your diagrams. Your AI draws; Zindex stores, lays out and renders them with deterministic code. Download any version as SVG, PNG, Mermaid or JSON, or your whole history in one file.