> ## Documentation Index > Fetch the complete documentation index at: https://info.bundle.social/llms.txt > Use this file to discover all available pages before exploring further. # Testing & Execution Logs > Test a flow on a real conversation or comment, and read the log of every run. Two tools for finding out what your flows do: a **test run** fires a flow on demand, and the **execution log** records every run, triggered or tested. ## Test A Flow **Endpoint:** `POST /api/v1/automations/:id/test` Runs the flow **once, synchronously**, and returns the finished execution. It uses the draft if there is one, otherwise the published version, so you can test changes before publishing them. A test skips the trigger entirely: keywords, `payloads`, `senderExternalIds`, the delay and the loop brake are not applied. It goes straight to the steps (conditions still run). Use it to check what the steps do, and a real comment or DM to check what triggers them. ```json theme={null} { "input": { "conversationId": "conv_abc", "text": "how much is the pro plan?" } } ``` Give it the context the steps need: * DM and click flows: `conversationId` of a real conversation in the team (plus `text` or `payload` if your conditions look at them), * comment flows: `fetchedCommentId` of a real comment (from the [Comments API](/api-reference/comments)), * flows with `SENDER_ID` conditions: pass the sender's ID in `input` yourself, a test doesn't take it from the conversation. **Tests are real.** Steps really send messages to the real person in that conversation, and the run counts toward your monthly quota. Test on a conversation with your own second account. Before anything runs, a test returns: * `400` if the flow has neither a draft nor a published version, or its definition doesn't pass validation (for example missing permissions), * `404` if the flow, conversation or comment isn't found in the flow's team, * `403` if the account's monthly run budget is used up, * `429` if the platform is rate-limiting the account right now (try again shortly). Once the steps run, the result comes back as `200` with the execution, even when it didn't go well: check its `status` and `error`. In the execution, `input.type` is `MANUAL_TEST`. *** ## Executions Every run is logged as an **execution**. This is where you answer "why didn't my automation reply?". Logs are kept for [30 days](/api-reference/data-retention) after the run. **Endpoints:** * `GET /api/v1/automations/:id/executions?teamId=team_123` - runs of one flow, newest first * `GET /api/v1/automation-executions/:id` - one run, with every step Filters for the list: `status`, `platform`, `socialAccountId`, `conversationId`, `fetchedCommentId`, `flowVersionId`, `triggerId`. Paginate with `limit` (max 500) and either `offset` or `cursor` (pass back `nextCursor`), same as [conversations](/api-reference/direct-messages/conversations#pagination). ```json theme={null} { "id": "exec_1", "teamId": "team_123", "flowId": "flow_abc", "flowVersionId": "ver_3", "triggerId": "trg_1", "platform": "INSTAGRAM", "socialAccountId": "sa_789", "fetchedCommentId": "fetched_1", "conversationId": null, "conversationMessageId": null, "senderExternalId": "1789000000000000", "currentStepId": "public-reply-2", "status": "COMPLETED", "input": { "type": "COMMENT_CREATED", "platform": "INSTAGRAM", "postId": "post_123", "text": "LINK please!" }, "output": { "completedAtStepId": "public-reply-2" }, "error": null, "startedAt": "2026-09-26T09:30:01.000Z", "completedAt": "2026-09-26T09:30:02.400Z", "createdAt": "2026-09-26T09:30:00.500Z", "updatedAt": "2026-09-26T09:30:02.400Z", "steps": [ { "id": "stepexec_1", "stepId": "private-reply-1", "status": "COMPLETED", "attemptCount": 1, "output": { "conversationId": "conv_abc", "messageId": "msg_42" }, "error": null, "startedAt": "2026-09-26T09:30:01.100Z", "completedAt": "2026-09-26T09:30:01.900Z" }, { "id": "stepexec_2", "stepId": "public-reply-2", "status": "COMPLETED", "attemptCount": 1, "error": null, "startedAt": "2026-09-26T09:30:01.950Z", "completedAt": "2026-09-26T09:30:02.400Z" } ] } ``` The execution also carries the related `flow`, `flowVersion`, `conversation`, `conversationMessage`, `fetchedComment` and `socialAccount` objects, so you rarely need a second request. ### Execution Status | Status | Meaning | | :- | :- | | `QUEUED` | Waiting to run (delay, or a platform rate limit) | | `RUNNING` | Steps are running | | `COMPLETED` | All steps done | | `SKIPPED` | Stopped on purpose. A condition didn't match (then `error` is empty and `output.skippedAtStepId` names the condition), or a limit kicked in (then `error.code` says which). | | `FAILED` | A step failed. The failing step has the details in its `error`. | | `CANCELED` | The flow was paused, archived or deleted while the run was still waiting | Each step has its own status: `PENDING`, `RUNNING`, `COMPLETED`, `FAILED` or `SKIPPED`. ### Error Codes Worth Knowing | `error.code` | Status | Meaning | | :- | :- | :- | | `AUTOMATION_MONTHLY_LIMIT_REACHED` | `SKIPPED` | Monthly run budget for this account is used up | | `AUTOMATION_CONVERSATION_COOLDOWN` | `SKIPPED` | DMs: this person got an automated reply less than 60 seconds ago | | `AUTOMATION_CONVERSATION_HOURLY_LIMIT` | `SKIPPED` | DMs: this person already got 6 automated replies in the last hour | | `AUTOMATION_COMMENTER_COOLDOWN` | `SKIPPED` | TikTok comments: this commenter got an automated reply less than 60 seconds ago | | `AUTOMATION_COMMENTER_HOURLY_LIMIT` | `SKIPPED` | TikTok comments: this commenter already got 6 automated replies in the last hour | | `AUTOMATION_SHARED_PAGE_CLAIMED` | `SKIPPED` | Another team with the same page connected already answered this comment or conversation | | `AUTOMATION_EXECUTION_DISABLED` | `SKIPPED` | Automations were briefly paused on our side (for example during a platform incident). Mostly seen on test runs: events that arrive during a pause don't create runs at all. | | `AUTOMATIONS_DISABLED` | `SKIPPED` | Automations were switched off for your organization before the run started | | `AUTOMATION_EXECUTION_EXPIRED` | `FAILED` | The run couldn't finish within an hour of its scheduled time (usually a long platform rate limit) | | `AUTOMATION_EXECUTION_TIMEOUT` | `FAILED` | The run took longer than 60 seconds | | `AUTOMATION_EXECUTION_STALE` | `FAILED` | The run was interrupted and didn't finish. Check which steps completed before acting on it. | Failed steps carry a readable `message`. Some come from the platform (for example, Meta rejecting a message), others from our own checks before sending (for example, a Facebook private reply to a comment older than 7 days, messaging not enabled for the account, or the monthly message limit). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.