io.datadef/mcp
io.datadef/mcp · 1.0.0
Generate, edit, and export data-architecture diagrams from your AI. Column lineage, PNG in chat.
Nobody here has read this server's code, because it publishes none. The description above is the maker's own, from the registry. A connection test shows whether it answers and what tools it says it has; it never calls a tool, so it cannot show what one does with your data.
Where it sits in the directory's order
What can be seen from outside, weighed as the directory publishes.
Observable signals: 40 of 100 points (at most 60 here)
| Input | Points | What was seen |
|---|---|---|
| Source published | held at 0 | No entry here publishes source: that is what puts it in this directory. |
| Licence stated | held at 0 | With no files, there is nothing for a licence to be stated in or held against. |
| Advisory state | held at 0 | Advisory databases index packages, and these entries publish none, so there is nothing to look up. |
| Auth declared | 15 of 15 | The registry entry declares a key or a token. |
| Latest connection test reached it | 0 of 20 | Not tested yet. |
Where it answers
https://datadef.io/mcpstreamable-httpAuthorizationrequired, secret: Bearer dd_live_... API key — create one at https://datadef.io/settings/mcp (7-day free trial includes MCP)
Connection test
The MCP handshake, then a request for the tool list, with no key and no data of yours. Run nightly, and by anyone, at most once every ten minutes per server.
It completed the handshake and listed 39 tools.
- Protocol
- 2025-06-18
- Calls itself
- datadef 1.1.0
- Handshake time
- 403 ms
- HTTP status
- 200
The tools it lists (39, as of 2 hours ago)
The tool-description rules found nothing in these descriptions. They look for instructions aimed at a model and for hidden characters; they cannot see what a tool does when it runs.
get_design_guideWhen: read once before your first canvas_* call, to build or to edit. Datadef's standard for a readable diagram (size, edge discipline, nested zones, notes) and the tool workflow: which canvas_* tool to use when, batching, laying out once, checking with canvas_view. It is the same standard Datadef's own generator follows, so a diagram you build by hand comes out indistinguishable from a generated one. Works without a subscription. Next: create_blank_diagram to build, or canvas_view to edit an existing diagram.
canvas_describe_canvasWhen: Use when you want a flat list of every node, edge and zone. canvas_view shows the same ids as a zone tree with positions and defects, and is usually the better first read. Get an overview of the current canvas: zones as a nested tree in reading order, every node with its id, type, icon and family, all connections with labels, and every note in full. Call this first, before any edit, so you work from real node ids instead of guesses. Operates on one Datadef diagram, named by diagram_id. If you have not already, call get_design_guide first: it carries Datadef's size, edge and zone stand
canvas_inspect_nodesWhen: Use before changing a node's fields or columns, to see everything it carries: columns, metadata, owners, neighbours. Get full detail for specific nodes: all columns with types and keys, metadata, schedule, formula, owners. Use before modifying a node so you preserve fields you are not changing. Batch: `nodes` (up to 10) takes a list; put every item in one call rather than one call per item. Operates on one Datadef diagram, named by diagram_id. Next: canvas_update_nodes, canvas_set_columns or canvas_replace_node.
canvas_search_iconsWhen: Use before adding nodes that should show a vendor logo (Snowflake, Kafka, Power BI) and you are unsure of the icon id. Find icon ids in Datadef's library (2000+ cloud and data-tool logos) by product name, e.g. "Amazon Aurora PostgreSQL", "k8s", "Azure Event Hubs". Pass `queries` to look up many names in one call (best id per name); pass `query` for the top matches of one name. Scores: 90+ sure, 75+ good. A name with no match has no icon; leave the icon field out rather than guessing. Batch: Pass `queries` with every vendor at once (up to 40, best id per name) instead of one call per ve
How this entry becomes a listing
For the maker. The catalogue lists what it can read, and this server publishes nothing to read yet. There are two routes, and only the first is open today.
Publish the source Open today
- Put the server's source in a public repository on github.com, with a licence.
- Keep a server.json in that repository naming this server,
io.datadef/mcp, and declare the repository in it:"repository": { "url": "https://github.com/owner/repo", "source": "github" }, adding"subfolder"when the server lives in a folder. - Publish that version to the official MCP registry.
- The nightly registry sweep records the declaration. This page stays, dated, and says the source is declared but not yet read.
- The catalogue reads declared repositories at a pinned commit, in batches run by hand, and lists the servers that meet the batch's rules (among them a server.json at that commit naming the server, an https endpoint and a licence). When it lists this server, this page links to the listing. There is no schedule, so no date can be promised.
List it through the maker studio Not open yet
From the official MCP registry, last updated there 64 days ago. The registry entry · datadef.io