I connect Claude or the API
In about half an hour you work with your diagrams from Claude Code and from your own scripts. Claude reads your models and prepares proposals; the API reads, checks and proposes. Neither ever changes a model without a person.
This path is for you if
You want Claude, a script or a pipeline to read your architecture, run iyedy's checks or prepare changes, without opening the editor.
Not for you if you want to draw diagrams by hand: take I’m new to iyedy.
At the end you have
- iyedy connected to Claude Code, answering with your own diagrams.
- A clear view of the MCP tools, and of how a proposal from Claude reaches the people who apply it.
- A token with the right scopes, and a first API call that works.
The steps
- Connect iyedy to Claude or Use iyedy in Claude CodeAdd iyedy to Claude Code and sign in once.
- MCP tools referenceThe 24 tools: 22 that read, 2 that store a proposal.
- How proposals reach your teamHow a proposal from Claude waits in iyedy until a person applies or rejects it.
- Your first API callRead your diagrams from a script, in a few minutes.
- Create and manage API tokensPersonal tokens and their lifetime; service accounts for CI.
- ScopesGive a token only what it needs: read, graph, propose.
- Rate limits and errorsRequests per minute by plan, and what each error means.
These docs publish a plain-text index for language models at /llms.txt. Give it to your assistant so it finds the right page.
Common questions
- What Claude can do with iyedy: what Claude can and cannot do with your diagrams.
- Assistants connected through MCP: what an AI client connected through MCP can read.
- Claude troubleshooting: Claude asks you to sign in again, or a diagram is missing.
Other paths
I’m new to iyedy
I was invited to review
I run an organisation
Was this page helpful?Your answer is anonymous and counts only if you accepted measurement.
What went wrong?
Don't write personal data.