🔑 Connecting Claude And Choosing Your Models

The plugin does not ship with AI credit. You supply your own Anthropic API key and it is stored on your own site. Nothing AI related runs until a key beginning sk-ant is saved.

Getting a key:

  1. Open console.anthropic.com/settings/keys and sign in. The AI tab has a Get Key button that opens the same page.
  2. Click Create Key, name it, and copy the value that starts with sk-ant.
  3. Paste it into Settings, AI and save.

The three model slots:

  • Primary, used for draft replies and for the resolver's answers. Default claude-sonnet-5.
  • Triage, the fast and cheap one used to classify each incoming ticket. Default claude-haiku-4-5.
  • Hard, reserved for complex work. Default claude-opus-5.

These are free text fields, so you can point them at any model id your key can call.

How calls are made:

Requests go to the Anthropic Messages API with prompt caching switched on, which is why the reusable parts of a prompt, such as your knowledge base excerpts, are sent as cached blocks. No sampling parameters are sent, because current Claude models reject them. Each call also states whether the model should reason before answering, so a short reply is never cut off by reasoning it did not need.

Every call is logged with the model, token counts, cache read and write tokens, latency and any error message. That log is what powers the spend estimate.

Budget control:

Monthly AI Spend Cap defaults to 50 USD. Before each call the plugin adds up this month's logged tokens, applies a blended rate and compares it to your cap. Once the estimate reaches the cap, AI calls stop and return a message telling you to raise it. The current estimate is printed under the field. Treat it as an estimate, not a bill: it uses one blended rate rather than per model pricing.

Ready to stop reading and start using?

Every plugin documented here works alongside the others, managed from one ResultsAIx Hub.