Skip to Content
Tares

Should I wire the AI SRE demo through the console or with the catalog file, and can I do both?

Tares gives you two separate ways to wire the AI SRE demo, and you pick one. The console path starts a plain tares up and uses the AI SRE demo template under Projects. The catalog path starts Tares seeded with TARES_CATALOG and only seeds an empty instance, so use a fresh data directory if you already ran tares up. Both end in the same wiring: three sources, one trigger and one agent.

How it works

Both paths create the same objects for the AI SRE guide, so the choice is about how you want to get there.

  • Console path. You start a plain tares up and do the rest in the browser. Under Projects you pick the AI SRE demo template, tagged demo. The setup page checks what Docker is running and fills in the Prometheus URL, the api-server URL and the log container from the actual containers and their published ports. If the demo stack is not up yet, the page shows the two commands to start it, with copy buttons.
  • Catalog path. You download the demo catalog file, export your Anthropic key and start Tares with TARES_CATALOG pointing at the file. Tares imports the catalog on start, while the instance is still empty. That is the only time it seeds, so a data directory that already holds objects is not seeded.

The two paths are not two halves of one setup. Each one builds the whole wiring on its own:

  • three sources keyed by the same service label: demo_metrics, demo_logs and demo_alerts
  • one incident trigger
  • one Tares agent, incident-first-look

If you do mix them, the project adopts an already imported demo catalog rather than duplicating it. Even so, both paths work best on a fresh data directory.

Set it up

Start the same way for both: run Docker, install Tares with uv or pipx, and bring up the demo system you will monitor. You need an Anthropic API key so the built-in agent can run.

  1. Start the demo stack and check that it is alive:
curl -O https://raw.githubusercontent.com/glassflow/tares/main/demo/docker-compose.yml
docker compose up -d
  1. Install Tares:
uv tool install tares          # or: brew install glassflow/tap/tares
  1. Choose one path.

Console path. Start Tares:

tares up

Open http://127.0.0.1:8787, go to Projects and pick the AI SRE demo template. Paste an Anthropic key on the setup page, or add one under Settings, Model providers, then click Start. Each object appears on its own page with a “part of project” badge. The project page also gets a Cause an incident button.

Catalog path. Download the catalog, export your key and start Tares seeded with it:

curl -O https://raw.githubusercontent.com/glassflow/tares/main/demo/catalog.demo.yaml
export ANTHROPIC_API_KEY=sk-ant-…
TARES_CATALOG=catalog.demo.yaml tares up
  1. If you already ran tares up in your current data directory, start on a fresh one instead, because the catalog only seeds an empty instance:
TARES_CATALOG=catalog.demo.yaml tares up --data-dir ~/tares-sre
  1. Start the MCP endpoint in a second shell with tares mcp, point your client at it, and cause an incident to watch the trigger wake the agent.

What you get

After either path you have the same working AI SRE:

  • One timeline per service, because all three sources carry service. Open Explore and pick api-server to see metrics, logs and alerts on one clock.
  • An incident trigger that fires when an alert event arrives. Prometheus keeps owning alerting, and Tares correlates what fired and wakes the agent.
  • An incident-first-look agent that writes a finding onto the timeline: what broke, since when, the likely cause tied to evidence, and a next action.

The path you pick changes how you get there:

  • Console path: the setup page fills in the URLs from your running containers, and you get the Cause an incident button on the project page.
  • Catalog path: the wiring lives in one file, so you can open it next to the console and find each piece. Starting from a fresh data directory keeps the result predictable.

To clean up, delete the data directory you ran with, or the --data-dir you passed.

When something else fits

  • Pick the console if you want the setup page to read your running Docker containers and fill in the URLs for you.
  • Pick the catalog file if you prefer one file you can read and start from. Use a fresh data directory with --data-dir if you already ran tares up.
  • Once the demo works, move on to your real systems. That step recreates the three sources against your stack and adds deploy events through a webhook source, as the AI SRE guide describes.

Still stuck? Get help

Last updated on
Quickstart