S.E.N.T.R.I

How it works

Give S.E.N.T.R.I a goal in plain language. It breaks the goal into steps, calls the appropriate skills — run PowerShell, edit code, drive a window or a browser, deploy to the cloud — checks its own results and reports back, asking for approval where you require it. You bring the model; S.E.N.T.R.I brings the skills, the memory and the guardrails.

Work of more than a couple of steps gets a checklist of phases that ticks across as the agent goes, so you can see how far along it is instead of watching a spinner. For bigger changes, switch on plan mode and it proposes a plan for you to approve before it touches anything.

What it comes with

• Agent Workspace — a main chat plus as many agents beside it as the job needs, with every session saved and restored on startup.
• Skills & Guardrails — built-in skills from PowerShell and code edits to Computer Use, Browser Use and Cloudflare, each switched on and gated one by one.
• Memory — what it learns about you and your estate, kept between conversations and filed per work folder.
• Chat History — every conversation restored on startup, searchable in the Logs tab and written to disk for audit.
• Voice — speak your prompt with local speech-to-text and hear the answer read back.
• Chat Integrations — reach your agents from WhatsApp, Slack and Discord.
• Scheduled prompts — put the agent on the clock and let it run unattended.

Bring your own model

Choose the provider that fits your policy and budget:

OpenRouter

Bring your own OpenRouter API key (BYOK) and pick any model OpenRouter offers. ServerEngine shows your remaining budget in the chat so you always know your spend.

1. 1

   Get a key

   Create an API key at openrouter.ai and paste it into the S.E.N.T.R.I settings.
2. 2

   Pick a model

   Browse the catalog filtered to Recommended, Most popular or All models. Every entry shows its context size, its price and its speed in tokens per second.
3. 3

   Star the ones you use

   Mark your go-to models as favorites, then filter the list to show only those — no scrolling past hundreds of models to find the two you actually run.

While you work, the budget meter in the chat header shows your spend against the monthly cap and how long until it resets.

Claude Code

Run S.E.N.T.R.I on Claude through the Claude Code Integration — ServerEngine drives the Claude Code CLI in the background as the model. There's no separate API key: it reuses your Claude Code login. Because the CLI is the backend, whatever it's configured for works out of the box — a Claude subscription (Pro/Max), an API key, or enterprise routes like Amazon Bedrock and LiteLLM gateway configs. No extra setup in ServerEngine.

Claude Code keeps its own tools — Read, Write, Edit, Glob, Grep, WebSearch and WebFetch — next to S.E.N.T.R.I's skills, sees the images you paste into the chat, and supports plan mode.

Codex CLI

The Codex CLI Integration does the same for OpenAI's models: ServerEngine drives the Codex CLI in the background and reuses its sign-in, so your existing subscription covers it and there is no key to paste into ServerEngine.

Ollama (fully local)

Point S.E.N.T.R.I at a local Ollama endpoint (for example http://localhost:11434) and pick a model to keep everything on-premises — no data leaves your network.

MCP servers

Settings includes an MCP integration for the Claude Code and Codex providers, so the MCP servers you already rely on can be connected to those CLIs from inside ServerEngine.

Switching model or provider

• Swap the model in the middle of a conversation without losing its history. If your model does not exist on the provider you switch to, a default one is picked for you.
• The main agent can hold chats on more than one provider at a time.
• Reasoning effort is set per chat from the model chip in the prompt line: Off, Low, Medium, High or xHigh.
• Whichever provider you use, the agent works from the same memory.

Interactive and unattended

S.E.N.T.R.I runs two ways, and the difference matters for your guardrails:

• Interactive — you chat with the agent in the console, or message it from your phone. Skills marked approval-required pause and wait for you to Accept or Decline, in the chat or by reply on WhatsApp.
• Unattended — a scheduled prompt or an API-triggered Prompt job, executed headlessly by the automation service. There is no one there to answer a confirmation, so the approval moves earlier: the scope and the skills that run is allowed are decided when you write it, and they are the boundary.

Both write the full conversation and every action to the Logs, which keep their own searchable S.E.N.T.R.I history.

In the chat

• Live checklist — multistep work shows its phases and ticks each one off as it finishes.
• Skill activity — every skill call shows as it runs, with the skill's name highlighted; the internal skill-call JSON is hidden from the transcript.
• Code lands in files — S.E.N.T.R.I writes code to disk with the Coder skill instead of pasting code blocks into the chat, and every edit shows as a red and green diff.
• Images — paste or drag PNG, JPG or WebP images into the prompt line. If the chat model cannot see images itself, the Vision skill reads them for it.
• Status — a thinking indicator with a live token count, a context readout, and for OpenRouter your remaining budget.
• Your language — S.E.N.T.R.I answers in the language you write in, and the app itself is available in 15 languages.

Next: the Agent Workspace — main chat, agents side by side and plan mode →