Learnings
What the agent learned, kept as create/remove facts rather than a decaying score.
Base path /api/v1/learnings on https://moabcrm.com. Authenticate with Authorization: Bearer $DASH_API_KEY and name the project with the Project-ID header. Writes are validated in declared mode; an operation this resource does not declare answers 405 operation_not_supported.
List learnings
GET /api/v1/learnings
curl "https://moabcrm.com/api/v1/learnings?verdict=like" \
-H "Authorization: Bearer $DASH_API_KEY" \
-H "Project-ID: pr_zest"Show me the learnings filtered by subject typedata_query {
"collection": "learnings",
"where": {
"verdict": "like"
}
}Returns a paginated list of learning objects, newest first.
{
"object": "list",
"data": [
{
"object": "learning",
"id": "obj_123",
"learning": "learning",
"body": "body",
"source": "source",
"created_at": "2026-07-25T12:00:00.000Z",
"updated_at": "2026-07-25T12:00:00.000Z"
}
],
"has_more": false,
"next_cursor": null
}Parameters
| Parameter | Type | Kind | Description |
|---|---|---|---|
subject_type | string | filter | The kind of record the belief is about (a collection key, or project). |
subject_id | string | filter | The specific record the belief is about, when it is about one. |
source | string | filter | Which pass or event produced the learning (e.g. prospect-agent, outbound-sync — historical rows may carry review-drafts). |
run_id | string | filter | The agent_runs row whose run produced this learning. |
verdict | string | filter | The decision this learning records, when it records one — a High Desert swipe (source: game:high-desert). One of like, nope, later. |
q | string | search | Free-text search across learning, body. |
sort | string | sort | One of created_at; prefix with - to reverse. |
limit | number | page | Rows to return, 1–100. Defaults to 20. |
cursor | string | page | next_cursor from the previous page. |
Retrieve a learning
GET /api/v1/learnings/draft-001
curl "https://moabcrm.com/api/v1/learnings/draft-001" \
-H "Authorization: Bearer $DASH_API_KEY" \
-H "Project-ID: pr_zest"Open the learning {id}data_get {
"collection": "learnings",
"object_id": "draft-001"
}Returns the learning object.
{
"object": "learning",
"id": "obj_123",
"learning": "learning",
"body": "body",
"source": "source",
"created_at": "2026-07-25T12:00:00.000Z",
"updated_at": "2026-07-25T12:00:00.000Z"
}Parameters
| Parameter | Type | Kind | Description |
|---|---|---|---|
id | string | path | The learning's id or slug. |
What comes back (read-only)
Every learning carries the envelope below. The fields under it are computed or stamped by the platform — they are returned, never sent.
| Field | Type | Description |
|---|---|---|
object | string | Always learning. |
id | string | Stable object id. Also accepted in place of slug on any path. |
slug | string | URL-safe name, unique within the collection. |
created_at | string | ISO-8601 timestamp of the first write. |
updated_at | string | ISO-8601 timestamp of the most recent write. The cursor sorts on this. |
verdict | string | The decision this learning records, when it records one — a High Desert swipe (source: game:high-desert). |
subject_type | string | The kind of record the belief is about (a collection key, or project). |
subject_id | string | The specific record the belief is about, when it is about one. |
links | object | Structured references to the records that evidence the belief, rendered as entity chips. |
run_id | string | The agent_runs row whose run produced this learning. |
metadata | object | Free-form provenance the producing pass attached. |
Errors
Failures use { error: { type, code, message, param } }. The full code table is generated at Error Codes.