Conversation V2

Conversation V2 uses explicit Conversation and ConversationReply schemas for discussions attached to metadata entities and fields. Breaking change: Clients that used /v1/feed must migrate. Human conversations now use /v1/conversations. Announcements, tasks, and generated activity use /v1/announcements, /v1/tasks, and /v1/activity. Update old feed clients to the matching resource endpoint and response schema.

Endpoints

Method	Path	Description
`GET`	`/v1/conversations`	List conversations.
`POST`	`/v1/conversations`	Create a conversation with `message` and `about`.
`GET`	`/v1/conversations/{id}`	Get one conversation.
`PATCH`	`/v1/conversations/{id}`	Apply a JavaScript Object Notation (JSON) Patch defined by Request for Comments (RFC) 6902.
`DELETE`	`/v1/conversations/{id}`	Delete a conversation.
`PUT`	`/v1/conversations/{id}/reaction/{reactionType}`	Add the caller’s reaction.
`DELETE`	`/v1/conversations/{id}/reaction/{reactionType}`	Remove the caller’s reaction.
`GET`	`/v1/conversations/{id}/replies`	List replies.
`POST`	`/v1/conversations/{id}/replies`	Add a reply.
`PATCH`	`/v1/conversations/{id}/replies/{replyId}`	Patch a reply.
`DELETE`	`/v1/conversations/{id}/replies/{replyId}`	Delete a reply.
`PUT`	`/v1/conversations/{id}/replies/{replyId}/reaction/{reactionType}`	Add a reply reaction.
`DELETE`	`/v1/conversations/{id}/replies/{replyId}/reaction/{reactionType}`	Remove a reply reaction.

List Parameters

Parameter	Type	Default	Description
`entityLink`	string	None	Restrict conversations to an entity or entity field.
`userId`	universally unique identifier (UUID)	None	Filter by user.
`filterType`	string	None	Filter by `OWNER`, `FOLLOWS`, `MENTIONS`, or `OWNER_OR_FOLLOWS`.
`resolved`	boolean	`false`	Filter by resolution state.
`startTs`, `endTs`	integer	None	Unix epoch millisecond time range. `startTs` cannot be greater than `endTs`.
`before`, `after`	string	None	Pagination cursors.
`limit`	integer	`10`	Page size from `1` through `100`.

Reply listing uses before, after, and limit. Its default limit is 20 and its maximum is 100. The {reactionType} path parameter accepts thumbsUp, thumbsDown, hooray, laugh, confused, heart, rocket, or eyes.

Create a Conversation

Send a POST request with message and about to create a conversation:

curl -X POST "{base_url}/api/v1/conversations" \
  -H "Authorization: Bearer {access_token}" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "Can we document the retention policy?",
    "about": "<#E::table::sample_data.ecommerce_db.shopify.orders>"
  }'

Both message and about are required. message is non-blank Markdown and about is an entity link.

Response Schemas

Conversation is the root schema. Its identity and source fields are id, source, and about. The resolved target appears in entityRef, with inherited domains in domains. Status fields cover timestamps, resolution state, and reply count. The schema can also include recent replies and reactions. User-created conversations contain the root message and createdBy. Activity-backed conversations identify the source event through activityEventId and activityTimestamp. ConversationReply identifies the reply, conversation, and author. It also contains the Markdown message, timestamps, and reactions. OpenMetadata stores replies separately and returns them through the reply endpoints.

Authorization

Listing requires ViewBasic on conversations. A request scoped with entityLink also requires ViewBasic on the target entity.
Creating a root conversation requires Create on conversations and ViewBasic on the target.
Reading a conversation requires ViewBasic on that conversation. OpenMetadata returns unauthorized direct reads as not found to avoid revealing the conversation’s existence.
Patching or deleting a conversation or reply requires the corresponding conversation permission evaluated against that resource. Editing message content is not implicitly allowed to every authenticated user.
Replying and reacting require permission to read the container and Create on conversations. Activity-backed conversations also require access to the activity target.

These checks replace the permissive behavior some callers encountered on the old feed API.

API Reference

Data Assets

Collaboration

AI SDK

SDK

Discovery

Lineage

Data Contracts

Teams And Users

Data Governance

Data Quality

Alerts

Incidents

Main Concepts

Conversation V2

Conversation V2

Endpoints

List Parameters

Create a Conversation

Response Schemas

Authorization

​Conversation V2

​Endpoints

​List Parameters

​Create a Conversation

​Response Schemas

​Authorization

Conversation V2

Endpoints

List Parameters

Create a Conversation

Response Schemas

Authorization