Setup & Models

The AI Agent needs an AI provider to talk to. Settings → AI Agent has a Model Provider toggle: Anthropic (Claude models, the default) or OpenAI-compatible, which covers Ollama, OpenAI, OpenRouter, or any other endpoint that speaks the OpenAI API. Pick whichever fits your needs, cost, and data-privacy requirements. You can switch providers at any time — your conversation history stays intact.

Anthropic

Getting an API key

1. Go to console.anthropic.com and create an account (or sign in).

2. Navigate to API Keys and click Create Key.

3. Give the key a name (e.g. s7-plc-buddy) and copy it — you won’t be able to see it again.

4. Add a payment method under Billing if you haven’t already. Anthropic uses pay-as-you-go pricing; typical usage in S7 PLC Buddy costs a few cents per session.

Entering your API key

1. Open Settings → AI Agent.

2. Make sure Anthropic is selected in the Model Provider toggle.

3. Paste your API key into the API Key field and click Save. The key is encrypted and stored locally using the Windows Data Protection API (DPAPI), and is never logged or transmitted to S7 PLC Buddy’s servers.

Choosing a model

Model	Best for	Extended thinking
Claude Opus 5 (recommended)	Complex reasoning, multi-step diagnostics, code explanation	Yes
Claude Sonnet 5 (default)	Faster responses, everyday questions, lower cost	Yes
Claude Haiku 4.5	Quickest and cheapest, simple lookups	No

A Custom… option in the model dropdown lets you enter any other model ID directly, in case a newer model isn’t listed yet. See the in-app Model list & pricing link for Anthropic’s current lineup.

Tip

Claude Opus 5 gets the most out of extended thinking for complex, multi-step fault scenarios where you want the agent to reason step by step before answering.

OpenAI-compatible (Ollama, OpenAI, OpenRouter)

Use this option if you want to run models locally for data privacy, use OpenAI’s own models directly, or access a wide range of hosted models (including open-weight ones) through OpenRouter with a single API key.

Ollama (local, on your own machine or network)

Nothing leaves your network with this option — the model runs entirely on hardware you control.

1. Install Ollama and pull a model that supports tool calling, e.g. ollama pull llama3.1.

2. In S7 PLC Buddy, open Settings → AI Agent and select OpenAI-compatible.

3. Click the Ollama preset button to fill in the Base URL (http://localhost:11434/v1, or your Ollama host’s address if it’s running on another machine on your network).

4. Leave API Key blank — Ollama doesn’t require one.

5. Enter the model name exactly as it appears in ollama list, e.g. llama3.1, and click Save Model.

Caution

Tool-calling reliability varies a lot with local and smaller models. Larger, more capable local models generally do better with the multi-step tool chains the AI Agent relies on. Anthropic or a hosted provider remains the most reliable choice for complex diagnostics.

OpenAI

1. Get an API key from platform.openai.com.

2. In Settings → AI Agent, select OpenAI-compatible, then click the OpenAI preset button to fill in the Base URL.

3. Paste your API key and click Save.

4. Enter a model name, e.g. gpt-4o-mini or gpt-5.6-luna, and click Save Model.

OpenRouter

OpenRouter gives you access to many providers’ models, including open-weight ones, through a single API key and endpoint.

1. Get an API key from openrouter.ai.

2. In Settings → AI Agent, select OpenAI-compatible, then click the OpenRouter preset button to fill in the Base URL.

3. Paste your API key and click Save.

4. Enter a model ID exactly as OpenRouter lists it, e.g. openai/gpt-4o-mini or meta-llama/llama-3.3-70b-instruct, and click Save Model.

Any other OpenAI-compatible endpoint

If you run your own inference server that speaks the OpenAI API, just enter its Base URL directly instead of using a preset button — everything else works the same way.

Reasoning display

When the model and provider support it, the agent’s reasoning summary shows in the same collapsible “thinking” view used for Anthropic’s extended thinking. This isn’t available on every model — S7 PLC Buddy detects support automatically per model, no configuration needed.

Limitations on this path

• Extended thinking and prompt caching (Anthropic-specific features) aren’t available here.
• No curated model dropdown — you enter the model ID as free text, since the available models vary by provider.
• There’s no reliable way to know an arbitrary model’s real output token cap. Max Output Tokens in Settings defaults to 4096; raise it if your model supports more and your responses are getting cut off.

Workspace folder

The AI Agent can read and write files in a workspace folder on your PC — useful for saving SCL migration output, exported reports, or reference documents the agent can consult. This works the same way regardless of which AI provider you choose.

Set the workspace folder in Settings → AI Agent → Workspace. The default is ~/Documents/S7 PLC Buddy.

The agent can:

• Read files you place in the workspace (e.g. datasheet PDFs, SCL source files)
• Write output files (migration results, structured summaries)
• List what’s in the folder

Caution

The agent can read any file in the workspace folder and its subdirectories. Do not put files with sensitive credentials in this folder.