Esta página aún no está en su idioma. Está leyendo el texto en inglés.
Automation
A built-in AI assistant
- 3
- AI services to choose from
- 60
- steps per answer by default
- 10 profiles
- flow run without asking
- up to 20
- your own MCP servers
What you get
A task in words
The assistant opens a profile, goes to an address, reads the page, clicks and types.
Your service and your key
Anthropic, Google, or any service compatible with OpenAI.
Several conversations at once
Each has its own set of profiles and its own Stop button.
Asks before anything big
By default the assistant asks you before creating, changing or deleting profiles.
Passwords stay hidden from the model
The app types a profile password and a 2FA code itself, and the model only gets the word "typed".
Your own skills and MCP servers
Add text skills and up to 20 MCP servers of your own.
Your AI service
Pick the provider, the key and the model in the AI service block.
An answer runs up to 60 steps, and you can raise the limit to 1000.
| Provider | What you need |
|---|---|
| Anthropic | Your Anthropic API key; the address is filled in by itself |
| Your Google AI (Gemini API) key; the address is filled in by itself | |
| OpenAI-compatible | The address and key of any service with that API, a model on this computer included |
Questions and answers
- Which model should I pick?
- The one you already pay for. Type its name into the Model field.
- Does the model see my passwords?
- No. The app types a profile password and a 2FA code, and the model gets the word "typed".
- How is this different from the MCP server?
- The MCP server opens ARMANOS to an assistant that lives in another program. This assistant lives inside ARMANOS.