Can I ask Tares what happened during a cron job that stopped straight from the console?

Yes, Tares has Ask, an assistant built into the console where you type a question such as what happened when a cron job stopped, and it investigates your data and shows each tool call as it works. Open it under Ask in the sidebar or press ⌘K from anywhere. Ask needs a model provider on the instance, either ANTHROPIC_API_KEY in the daemon’s environment or one added under Settings, Model providers. It runs on the instance’s default provider, and the chat says which.

How it works

Ask runs as a chat loop on the daemon, with up to 10 tool rounds per question. Its tools are thin calls to the daemon’s own read endpoints, so it works from the same read API everything else uses.

For a cron job that stopped, these tools matter most:

• List sources and connectors, and describe any catalog object, including its schema, entities, freshness and samples.
• Read recent events from a source, so you can see the last things the job reported.
• Read a correlated timeline, across every source or only one project’s, so you can line up the job’s last events with what else happened around that time.
• Check live health and field coverage, which is how Ask answers questions like “is anything not ingesting?” and “what looks sparse?”.

The model is the default of your provider. On Anthropic that is the value of TARES_AGENT_MODEL, which is claude-sonnet-4-6 unless you set it.

Set it up

1. Start Tares locally and open the console. On Tares Cloud, open the console from your workspace.

   tares up

2. Give the instance a model provider. Pick one of two ways:

   • Put ANTHROPIC_API_KEY in the daemon’s environment.
   • Add a provider in the console under Settings → Model providers.

3. Open Ask in the sidebar, or press ⌘K from any page. The chat tells you which default provider it is running on.

4. Ask about the job in plain words, naming the service or source it reports to. For example: “What happened to the nightly backup job? When did its events stop?”

5. Watch the tool calls Ask makes. It lists the sources, reads recent events and reads the timeline, then gives its verdict. Follow up with narrower questions, such as what else happened in the same window.

6. If Ask says a source looks sparse or has stopped ingesting, ask it why a label is missing or what the field coverage looks like. The same read tools answer those questions from live health.

For Ask to explain a stopped job, the job’s logs or events need to be flowing into Tares as a source. Ask reads what is on the instance, so it can show you the point where the events stop and what surrounded it.

What you get

• An answer from inside the console, without wiring up an external client first.
• A visible investigation. Each tool call shows as Ask works, so you can check how it reached its verdict.
• Help locating the gap. Ask can tell you whether a source is still ingesting, how fresh it is and what its last events said.
• Context from other data. A correlated timeline lines the job’s last events up with other sources on the instance.
• Optional follow-up changes. Ask can draft label and trigger proposals as cards you apply or skip, for example a trigger to catch the next time the job goes quiet. The proposals change nothing until you apply them, and applying goes through the same validated management APIs the console forms use. A trigger applied from Ask goes to the Default project.

For the wider picture of using Tares to investigate incidents, see AI SRE.

When something else fits

• Ask reads this one instance and sees only what the API exposes. If the cron job never sent anything to Tares, there is nothing for it to read, so connect a source for the job first.
• For an agent in your own workflow, such as Claude Code, Cursor or your own service, connect it over MCP to get the same read surface plus subscriptions. The Ask page describes that choice.
• To set up a whole project from a goal instead of asking questions, use the guided setup.

Related