Publishing to Informatica IDMC

Publish your MCP servers, REST APIs and Agent Cards to Informatica Intelligent Data Management Cloud (IDMC). They appear in Data Governance and Catalog (CDGC) as catalog objects that data stewards can search, govern and trace.

What Gets Published

A²D publishes each asset as a small hierarchy of catalog objects. Input and output fields are separate objects, so CDGC shows which inputs flow to which outputs as lineage.

• Tool input fields come from the tool’s inputSchema.
• Tool output fields come from the tool’s outputSchema, or from its first JSON mock response when there is no output schema.
• Nested fields are named by their full path, so fields with the same name stay apart (for example totals.tax.amount and items[].unitPrice.amount). Array items are written as []. Only leaf fields are published; container objects like orders show up only in the field names and JSON paths.
• $refs to definitions in the same schema (#/$defs/..., #/definitions/...) are followed, so a shared address definition publishes billingAddress.city, shippingAddress.city and so on. A $ref that points outside the schema, or back to itself (a recursive structure), is published as a single field.
• oneOf / anyOf variants are merged: every field of every variant is published. A field is Required only when every variant requires it. For example, a payment that is a credit card, PayPal or bank transfer publishes payment.cvv, payment.paypalEmail and payment.iban.
• Field types come from type, or from const / enum values when there is no type (for example {"const": "paypal"} is a string). Fields with several types are written like string|integer.
• Limits — fields nested more than 8 levels deep (an array of objects counts as two levels), and fields after the first 500 of a schema, are not published. The publish report lists them.
• Lineage links an input field to the output field with the same path (for example order_id → order_id). If no output has the same path, it links to the one output with the same field name (for example status → order.status). When several outputs share that name, no link is made.
• Agent Skills have no field objects: an A2A skill takes a prompt and returns a reply, with no field schema.
• Provenance — every top-level object carries A2D Asset ID, A2D Published By, A2D Published At and A2D Lifecycle.

Disabled tools, endpoints and skills, and extended-only agent skills, are not published.

Prerequisites

• An MCP server, REST API or Agent Card in A²D
• Metadata Command Center and Data Governance and Catalog enabled in your IDMC organization
• An IDMC user that signs in with a username and password (A²D uses the IDMC login API, not single sign-on) and can:
  • In Metadata Command Center: create custom catalog source types, and import and publish custom metadata models (once per IDMC organization, for the A2D Agentic Assets model)
  • In Metadata Command Center: create, view, edit and run catalog sources, upload metadata files, and view their jobs (Monitor → Jobs)
  • In Data Governance and Catalog: search and view the published assets
• Your IDMC login URL, for example https://dm1-emce.informaticacloud.com

Your IDMC administrator grants these in Administrator → User Roles. A²D also lists them under Settings → User Settings → IDMC Credentials.

Configure Your IDMC Credentials

1. Go to Settings → User → IDMC Credentials
2. Enter the IDMC login URL, Username and Password
3. Click Test connection to check them
4. Click Save Credentials

Like your Anypoint credentials, IDMC credentials are stored in this browser only (local storage). A²D sends them to its server only while you publish and never saves them.

You can also enter them the first time you publish to IDMC.

Publish an Asset

1. Open the MCP server, REST API or Agent Card (or use Publish from the list page)
2. Click Publish and choose Informatica IDMC
3. Choose the target catalog source:
   • New catalog source — A²D creates a custom catalog source (default name A2D Agentic Assets) and owns its content
   • Existing catalog source — add the asset to a custom catalog source that already exists
4. Click Publish to IDMC

The dialog shows each step. Under the running step it shows what IDMC is doing, how long the step has been running and when A²D last checked. For IDMC jobs it also shows the job name and ID, so you can follow the job in Metadata Command Center (MCC) under Monitor → Jobs. Open Activity to see everything the publish has done so far.

The scan in IDMC usually takes 5–10 minutes. Keep the dialog open until it finishes; if you close it, open Publish → Informatica IDMC on the same asset again to continue.

When the publish finishes, find the objects in Data Governance and Catalog under the catalog source you chose.

Publish Report

When the publish succeeds, the dialog shows a Publish report. For each asset, it lists every tool or REST operation with:

• the input and output fields pushed (REST parameters count as inputs)
• the number of lineage links from inputs to outputs
• how many oneOf / anyOf schemas were merged
• the fields that were not published, or were published as one opaque field, and why: nested too deep, over the field limit, a $ref that could not be resolved, a recursive $ref, or a schema with no type

Click Download JSON to save the report. The report is also part of the publish run’s result in the REST API and the Platform MCP tools.

New vs Existing Catalog Sources

Only custom catalog sources that use uploaded CSV files can receive A²D assets. Other sources are listed as not supported.

A²D never deletes a catalog source.

Troubleshooting

Next Steps

• Publishing to Exchange — publish the same assets to Anypoint Exchange
• Bulk Publishing — publish several assets in one catalog source with one scan
• Structured Content (Output Schemas) — add output schemas so tool output fields are published
• Creating Agent Cards — describe skills that appear in the catalog

Your agentic assets are now governed alongside the rest of your data in Informatica IDMC.