> ## Documentation Index
> Fetch the complete documentation index at: https://docs.praxa.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Praxa API Playground endpoint coverage and test plan

> See every active Praxa API Playground operation, required scopes, success proof, negative tests, and the endpoint intentionally excluded from testing.

The API Playground covers all 15 active operations in the checked public
Execution Fabric and memory-candidate OpenAPI documents. It intentionally
excludes the pending public approval operation and the deployment-specific
Integration Gateway `/v8` surface.

## Active endpoint coverage

| Operation                                         | Scope                     | First success         | Verify before claiming success                                |
| ------------------------------------------------- | ------------------------- | --------------------- | ------------------------------------------------------------- |
| `POST /v1/execute`                                | `execute:write`           | `202`                 | Read or stream the returned run to terminal.                  |
| `GET /v1/runs/{id}`                               | `runs:read`               | `200`                 | Match run ID, status, and tenant-owned projection.            |
| `GET /v1/runs/{id}/events`                        | `runs:read`               | `200` SSE             | Persist sequence, resume, and observe terminal state.         |
| `POST /v1/runs/{id}/cancel`                       | `runs:write`              | `200`                 | Reconcile until terminal; request acceptance is not rollback. |
| `GET /v1/usage`                                   | `usage:read`              | `200`                 | Check range, UTC buckets, totals, and tenant isolation.       |
| `GET /v1/webhooks`                                | `runs:read`               | `200`                 | Require tenant-owned metadata and no signing secret.          |
| `POST /v1/webhooks`                               | `runs:read`, `runs:write` | `201`                 | Store the one-time secret and verify a signed delivery.       |
| `PATCH /v1/webhooks/{endpointId}`                 | `runs:read`, `runs:write` | `200`                 | Read back fields and exercise the new filter.                 |
| `DELETE /v1/webhooks/{endpointId}`                | `runs:read`, `runs:write` | `204`                 | Confirm disabled state and remove the stored secret.          |
| `GET /v1/webhook-deliveries`                      | `runs:read`               | `200`                 | Correlate endpoint, run, event, status, and cursor.           |
| `POST /v1/webhook-deliveries/{deliveryId}/replay` | `runs:read`, `runs:write` | `202`                 | Follow replay delivery to its final status.                   |
| `POST /v1/memory/records`                         | `memory:write`            | `201` or replay `200` | Query exact subject and prove replay identity.                |
| `POST /v1/memory/query`                           | `memory:read`             | `200`                 | Require lexical mode, filters, and subject isolation.         |
| `GET /v1/memory/export`                           | `memory:read`             | `200` NDJSON          | Parse all lines and follow the opaque cursor.                 |
| `DELETE /v1/memory/records/{recordId}`            | `memory:write`            | `200`                 | Confirm erasure, content-free receipt, and replay identity.   |

## Intentionally excluded endpoints

`POST /v1/runs/{id}/approve` remains hidden and pending in the checked
Execution Fabric specification. The playground does not expose a button for a
route that is not activated. Use the [approvals status guide](/fabric/api/approvals)
to understand the contract without sending a request.

The Integration Gateway `/v8` and remote MCP endpoint are deployment-specific.
Use that deployment's origin and delegated OAuth authority through the
[API Reference](/api-reference/overview) or [MCP setup guide](/mcp/quickstart).

## Run the minimum acceptance matrix

For every operation you adopt, save a test record containing the environment,
operation, synthetic tenant/subject, credential fingerprint—not the credential
itself—request ID, idempotency key fingerprint, response status, request ID,
readback result, cleanup result, and timestamp.

| Lane           | Required checks                                                                      |
| -------------- | ------------------------------------------------------------------------------------ |
| Contract       | Method, path, required fields, unknown-field rejection, response schema              |
| Authentication | Missing, malformed, revoked credential                                               |
| Authorization  | Correct scope, wrong scope, foreign resource, tenant and subject isolation           |
| Replay         | Exact retry, changed input under old key, timeout reconciliation                     |
| Availability   | `429`, retryable and non-retryable 5xx, bounded timeout                              |
| Lifecycle      | Run terminal state, event resume, webhook final delivery, or memory deletion receipt |
| Cleanup        | Revoke key, disable endpoint, erase test candidate, remove provider fixture          |

<CardGroup cols={2}>
  <Card title="Test your integration" icon="flask" href="/tutorials/test-your-integration">
    Build the complete positive, denial, replay, isolation, degraded-state,
    and cleanup harness.
  </Card>

  <Card title="Troubleshoot failures" icon="triangle-exclamation" href="/api-playground/errors">
    Map each typed problem to a safe retry, recovery, or stop decision.
  </Card>
</CardGroup>