List agent conversation task records

Read task projections in an accessible agent conversation to recover call states, display content, and artifacts. This API does not send new messages or execute or retry tasks.

GET https://api.moi.matrixorigin.cn/v5/workspaces/{workspace_id}/conversations/{conversation_id}/tasks

Before you call this API

PRODUCT_API_BASE_URL uses the product API prefix https://api.moi.matrixorigin.cn/v5. For a dedicated deployment, use its corresponding product API address.

PRODUCT_API_BASE_URL includes the product prefix for the target environment. Prepare your own credentials and an authorized WORKSPACE_ID. The path ID must match X-Workspace-ID. Replace example IDs with values actually returned by that environment. Obtain conversation IDs from agent conversations creation, list, or detail APIs, not independent numeric LLM conversation IDs.

Request

curl --get "$PRODUCT_API_BASE_URL/workspaces/$WORKSPACE_ID/conversations/$CONVERSATION_ID/tasks" \
  -H "X-API-Key: $PRODUCT_API_KEY" \
  -H "X-Workspace-ID: $WORKSPACE_ID" \
  --data-urlencode "agent_id=$AGENT_ID" \
  --data-urlencode "agent_workspace_id=$AGENT_WORKSPACE_ID"

Query parameter

Type

Required

Description

agent_id

string

No

Validates the agent scope of the conversation.

agent_workspace_id

string

No

Agent source workspace: current workspace or system only. Defaults to the current workspace when agent_id is provided without this value.

This operation has no limit/offset/cursor; the current handler does not read these parameters. Regular conversations are restricted to their creator. System workflow-authoring conversations are scoped separately by workspace but still require identity and role authorization.

Successful response

{"code":0,"data":{"items":[{"kind":"task","id":"task-example","contextId":"conversation-example","status":{"state":"working","timestamp":"2026-11-12T01:02:03Z"},"metadata":{"eventCursor":3,"agentId":"agent-example"}}]}}

Field

Type

Description

data.items

array

Conversation task records projected as A2A Tasks. Empty array when there are no results.

data.items[].kind

string

task。

data.items[].id

string

Actual task ID; use this value for subsequent task queries.

data.items[].contextId

string

Agent conversation context ID, using A2A camelCase.

data.items[].status.state

string

A2A state, distinct from raw internal runtime state enums.

data.items[].status.timestamp

string

Task update time.

data.items[].status.message

object

Optional current display message; parts may contain several content types.

data.items[].artifacts

array

A2A artifacts returned when available; may be omitted.

data.items[].metadata

object

Task display metadata, potentially including eventCursor, manifestId, provider, agentId, and traceId.

A2A states include submitted, working, input-required, auth-required, completed, failed, canceled, and rejected. queued projects to submitted; internal states such as cancel_requested/reconciling project to working. Do not compare state enums directly across APIs.

Interpret the result

Results may include display content still changing during execution. A successful request does not mean all tasks are complete, nor does it return a raw private event log. Tasks are merged in conversation-turn persistence order; do not infer chronology from ID strings. Private events are not returned as a direct event array. Missing artifacts do not prove a task never ran.

For subsequent task queries or control, retain the current workspace, agent source workspace, agent ID, contextId, and actual id. Do not use contextId as a task ID or pass a task id to conversation message APIs.

Errors

The service validates conversation scope before reading task history. Missing or out-of-scope conversations/tasks may return 404; invalid parameters return 400 and permission failures 403. Corrupt task or candidate display data may return 500; dependency failures may return 503/504. Do not treat an empty list as success on failure or resend messages because a read failed.

Last updated on