Alpha software. Expect bugs. noHuman Team runs AI workers on your own machine with real file, browser and desktop access — start on work you can afford to have go wrong, and leave the approval queue on. What is rough right now

Back to home

Advanced Configuration

Start with the settings for the scope you want to change: the computer, the selected team, or one member. Team responsibilities and files stay separate; connected AI accounts and the computer's license are shared.

Responsibilities and personal instructions

Open a member's settings from the conversation's More menu. Use Responsibilities for the job, Reporting line for delegation, and Personal instructions for communication style and working preferences.

For example: “Give three headline options and explain which audience each fits.” Keep durable project facts in Memory or the team's files. Instructions guide the model; they are not a substitute for tool permissions or the member's Autonomy controls.

noHuman Team v4.2.2Explore the interface

Responsibilities

Give your team clear responsibilities and connected services. Changes here stay in this preview.

Clear responsibilities help your noHuman decide what to work on and when to involve others.

Give a noHuman clear responsibilities without changing its saved conversations.

Review the save or apply notice: some changes apply live, while others require a restart. For a different job rather than a personality adjustment, read role migration. A non-leader role change restarts the member; saved history and Memory are not automatically cleared.

Capabilities, skills, and connected tools

Use Tools to review the selected team's installed capabilities and catalog. Use a member's Capabilities settings to see its tools, plugins, and skills, or Connected tools for external tool connections.

Skills are bundles built around a SKILL.md manifest and may include scripts or supporting files. Their scope matters: some are shared across the team and others belong to one member. Read the bundle and its approval or restart notice before installing it.

A skill or tool can run commands or contact an external service. Only install content you trust. Store service credentials through AI accounts or team integrations, not in a shared instruction or skill file.

Schedules, reminders, and heartbeats

Ask a member to create a reminder or recurring task, then verify the schedule and next run. The computer and worker must be running for local scheduled work to execute. A request such as “check the inbox every morning” also needs a connected tool and appropriate permission.

noHuman Team v4.2.2Explore the interface

Reminders

See scheduled follow-ups that run at their due time and survive restarts. Pending reminders are still waiting; fired reminders already happened.

· Private memory

Review a noHuman’s scheduled follow-ups in Memory → Reminders.

Open the member's Schedule settings to enable or disable heartbeat checks, set an interval, choose save-state or proactive modes, and use the team's helper model or set a supported override. The page shows recent checks and whether the saved change was applied or needs a restart.

The helper model is one small model for the team's background jobs: heartbeat checks, the Archivist that saves to Memory, and loop checks. Choose it in Team settings → AI models → Helper model; Automatic uses the fastest small model your connected Claude or ChatGPT subscription offers, and Test now shows whether it answers. It is separate from each member's own model.

Proactive checks can use model tokens even when no visible message appears. Review the recorded checks and provider usage rather than assuming an idle team is always free. Set work limits and confirm recurring external actions before leaving them unattended.

Workspace, team data, and Memory

Each team has a shared workspace and its own local data. Members also have individual working locations. Use Team settings to inspect the actual paths; do not move internal files while workers are running.

Open Library for team files, saved Memory, and knowledge. The Memory button in a conversation opens that member's notes. Working context is separate: compaction summarizes it, and a reset starts a new session without deleting saved history, files, or Memory.

Back up both the workspace and team data, or use Team settings → Backups & recovery. Native workers and their terminals run with your operating system's permissions. Separate team folders organize work; they do not create a filesystem sandbox.

Terminal and diagnostics

Use the conversation's More menu to open Terminal or Activity & logs in the right panel. The terminal uses that member's local environment. Read a command before running it: it can change real files or services.

Never paste secrets into chat, shared files, screenshots, or support reports. If a command or install times out, inspect its status before repeating it. Read Troubleshooting for account, startup, and diagnostics checks.