# 11.1 Stop first, then read the record (/managing-ai-workers/when-the-worker-goes-wrong/stop-first) --- type: Document title: "11.1 Stop first, then read the record" description: "What to do first when an AI Worker's run goes wrong: stop it, read its record and check what changed, on Claude and ChatGPT." status: stable order: 211.1 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-10-08T13:21:13Z chapter: "11" part: II expert_status: required concepts: [ "11.1" ] last_verified: 2026-10-08 sources: - id: v1-workflow-design title: "Workflow Design & Diagnosis (Panaversity, first edition, verified 2026-10-08)" resource: https://agentfactory.panaversity.org/docs/workflow-design-diagnosis-crash-course - id: v1-problem-solving title: "Problem Solving with General Agents: A 90-Minute Crash Course (Panaversity, first edition, verified 2026-10-08)" resource: https://agentfactory.panaversity.org/docs/problem-solving-crash-course - id: anthropic-scheduled-tasks title: "Schedule recurring tasks in Claude Cowork (Claude Help Center, updated 6 October 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/13854387-schedule-recurring-tasks-in-claude-cowork - id: anthropic-cowork title: "Get started with Claude Cowork (Claude Help Center, updated 6 October 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/13345190-get-started-with-claude-cowork - id: openai-scheduled-tasks title: "Scheduled tasks in ChatGPT (OpenAI Help Center, verified 2026-10-08)" resource: https://help.openai.com/en/articles/10291617-scheduled-tasks-in-chatgpt - id: openai-work-codex title: "ChatGPT Work and Codex (OpenAI Help Center, verified 2026-10-08)" resource: https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex generated: at: 2026-10-08T13:21:13Z by: esl-rewrite/1.2.0+ksor.1 trust_tier: unverified build_id: sha256:2f1e14f2a0f11dbc37eb5856ca9b7d35b4ce4894da74c3328cf61a3c17c03861 dirty: true ksor_version: 0.0.60 --- **In everyday life.** Your washing machine floods the floor. You turn off the water before you look for the cause. And you look at where the water came from before you buy a new machine. **What you will learn.** What to do first when an AI Worker's run goes wrong: stop it, find out what it did, and check what really changed. You practice it in steps 1 and 2 of the lab. **Why it matters.** A quick fix destroys the evidence and lets the fault happen again. In this chapter's story, Brightline Wholesale Supply, the book's example company, uses an AP Worker, the AI Worker that handles the bills the company owes. On Friday, January 15, 2027, it drafted five replies to vendors. Each said that a remittance notice, the message that tells a vendor what was paid, was attached. Nothing was attached, and Maria, the office manager, sent all five. Your first reaction is to fix a failing worker at once. But that destroys evidence. Work in this order instead: 1. *Stop* the run. Pause any job that runs on a schedule, or that starts when something happens, until you know the cause. 2. *Inspect* the **task record** before you decide what went wrong.[^v1-problem-solving] 3. *Contain* the harm. Hold anything the run set in motion, such as a payment, and have a person do the job by hand. Remove risky access the job does not need: unused connectors to other apps, write access and the ability to send messages. 4. *Correct* the cause, once you know it (11.3 and 11.5). 5. *Verify* everything the run affected outside the conversation: messages, files and records.[^v1-workflow-design] 6. *Resume* the work with approvals on, so the worker asks before it acts. Give the worker back only the authority its Role Contract, its one-page job definition, already allows. Do it only once the worker's owner agrees the fix has worked. A **task record** (Chapter 6) is everything a worker leaves behind about one run. It shows what the worker was asked, the steps it took, what it read, what it could reach and what it produced. The record tells you what the worker reports it did. Only the connected systems tell you what actually changed. If the record says "email sent," open the sent folder. Compare the record with the brief, the worker's written instructions, step by step. The Friday record showed the fault at once. Its step 2 said "access denied." Its step 3 said "drafting replies." A worker that cannot reach its input should stop and say so. This one guessed. ![The title reads "Stop first, then read the record." The line below it reads "Find the cause before changing the worker." Six numbered steps run in a row. Step 1, "Stop": "Pause the run and recurring job." Step 2, "Inspect", is highlighted: "Read the task record." Step 3, "Contain": "Restrict risky access and actions." Step 4, "Correct": "Fix the confirmed cause." Step 5, "Verify": "Check affected systems." Step 6, "Resume": "Approvals on. Stay within the Role Contract." Under step 2, a panel, "Inspect the task record," has five rows. Asked: "The brief or scheduled prompt." Did: "The steps taken, in order." Read: "The sources and files opened." Reached: "Connectors, tools and skills." Produced: "Outputs and their count." A second panel, "Check both sources of evidence," has two rows. Task record: "What the worker reports it did." Connected systems: "What actually changed." Under it, "Email sent" has an arrow to "check the sent folder." A banner at the bottom reads "Correct only once the cause is known. Resume only within existing authority."](img/stop-first.png) *Figure 11.1. Stop first, then read the record. Correct only once the cause is known.* > **Anthropic, as verified 8 October 2026.** Claude's *Scheduled* page lists a task's past runs, and each run is its own session. The page lets you pause, edit or run the task.[^anthropic-scheduled-tasks] A running task shows its progress at each step and delivers its outputs to the session.[^anthropic-cowork] There are more details in [The details on both AI vendors](both-ai-vendors.md). > **OpenAI, as verified 8 October 2026.** ChatGPT's *Scheduled* page shows results and lets you pause tasks. OpenAI's documentation gives the reasons a task pauses: it needs an approval or further action, it is inactive, or its chat was deleted. A quiet event-triggered task may be paused, may be waiting for approval, or may have lost its connection to its app.[^openai-scheduled-tasks] ChatGPT Work is an agent that acts across your apps and files. While one of its tasks runs, you can review its progress, change its direction and approve important actions. OpenAI's documentation describes no separate step log.[^openai-work-codex] There are more details in [The details on both AI vendors](both-ai-vendors.md). **Comparison.** Both AI vendors show you the results of past runs and let you pause a task. Only OpenAI's documentation lists the reasons a scheduled task pauses, so on OpenAI, start with that list. On Anthropic, start with the run itself. **What stays the same.** Stop before you fix. The record shows what the worker reports it did. The systems it worked on show what changed. Check both. [^v1-problem-solving]: Problem Solving with General Agents: A 90-Minute Crash Course, Panaversity, first edition. [^v1-workflow-design]: Workflow Design & Diagnosis, Panaversity, first edition. [^anthropic-scheduled-tasks]: Schedule recurring tasks in Claude Cowork, Claude Help Center. [^anthropic-cowork]: Get started with Claude Cowork, Claude Help Center. [^openai-scheduled-tasks]: Scheduled tasks in ChatGPT, OpenAI Help Center. [^openai-work-codex]: ChatGPT Work and Codex, OpenAI Help Center.