Can I ask Tares what happened during slow database queries straight from the console?

Yes, Tares lets you ask what happened during slow database queries from the console with Ask, once the instance has a model provider. Open Ask in the sidebar or press ⌘K, type your question, and watch it investigate your data step by step. Ask runs on the instance’s default provider, and the chat tells you which one.

How it works

Ask is the assistant built into the Tares console. It runs as a chat loop on the daemon, with up to 10 tool rounds per question. Each tool is a thin call to the daemon’s own read endpoints, so Ask sees exactly what the read API exposes and nothing more.

For a question about slow database queries, Ask can:

• List your sources and connectors, and describe any catalog object, including its schema, entities, freshness and samples.
• Profile a source’s fields, list entities and read recent events.
• Read a correlated timeline across every source, or only across one project’s sources.
• Check live health and field coverage, so it can tell you whether a source stopped ingesting or why a label is missing.

Every tool call shows in the chat while Ask works, so you can follow how it reaches its verdict. It reads what is already in your instance. If the logs, metrics or rows from around the slowdown flow into Tares, Ask can find and correlate them. If nothing from that window is ingested, add a source for it first, and Ask can then help you check that the data is arriving.

Ask needs a model provider on the instance. It uses the instance’s default provider, and the same providers power Tares agents. On Anthropic the model is TARES_AGENT_MODEL, which is claude-sonnet-4-6 unless you set it.

Set it up

1. Give the instance a model provider. Either set ANTHROPIC_API_KEY in the daemon’s environment before you start Tares, or add a provider in the console under Settings, then Model providers.

   tares up

2. Make sure the provider you want is the instance’s default. Ask runs on the default provider and does not let you pick another one per question.

3. Open Ask in the sidebar, or press ⌘K from anywhere in the console. The chat names the provider it is using.

4. Ask your question in plain words, naming the time window and the service or database you care about. For example:

   Which service had the most errors in the last 24 hours, and what happened to it?

   Adapt it to your case, such as asking what happened to the database service during the slow queries this morning.

5. Read the tool calls as they appear, then read the verdict. Follow up in the same chat to narrow the window, check a different source or look at a single entity.

6. If Ask proposes a label or a trigger, it shows a card. Choose Apply or skip it. The proposal changes nothing until you apply it. A trigger applied from Ask goes to the Default project.

For the wider picture of how Tares fits into incident work, see AI SRE. The full reference for the assistant is on the Ask page.

What you get

• An answer inside the console, with no external client to wire up first.
• A visible trail of tool calls, so you can check how Ask got to its conclusion.
• Debugging help for the data itself: whether a source is ingesting, what looks sparse, and why a label is missing.
• Draft labels and triggers as cards you apply or skip, so a slowdown you just investigated can become something Tares watches for next time. Apply goes through the same validated management APIs the console forms use.
• The same provider as your Tares agents, so you set up model access once.
• A starter-question empty state for exploring, debugging and organizing, which helps on an instance you have just connected.

When something else fits

• Ask lives inside the console and reads this one instance. If you want 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.
• If you want to set up a whole project from a goal instead of asking questions, use guided setup.

Related