Versions and upgrades

RoboAPI 0.4.5Copy Anchor Link

Released on October 29, 2026. Install from PyPI using the dependencies for your robot in the installation guide.

What's newCopy Anchor Link

• Select models from the service: enter a key during setup, retrieve available models, and choose one without typing a model ID.
• Connect before saving: in the console, Connect, select a model, then Save. Keep the current selection when available; changing the address or key reloads the list.
• Default production address: the Base URL defaults to https://api.pokebot.com/v1, with compatible custom services still supported.
• Retry failed connections: fetching the list does not save entered credentials or create a task session. Failure preserves saved settings and the current input.

New installations need network access to the model service. See installation and the local console. Hardware adapters, calibration, and action execution keep their existing workflows.

Upgrade an existing installationCopy Anchor Link

1. Select Robot Stop, then shut down RoboAPI. Stop any stdio process launched by an MCP client too.
2. Back up ~/.config/roboapi/ and activate the existing Python environment.
3. Choose the full command matching your robot from installation or Franka setup. Add --upgrade and pin ==0.4.5. For example, for FR3 system 5.9+:

source ~/roboapi/.venv/bin/activate
uv pip install --upgrade 'roboapi[franka,gripper]==0.4.5'

For FR3 system 5.6.0 and Panda, also retain the matching SDK build version and index URL.

4. Restart with the existing configuration: use roboapi http for the default location, or roboapi http -c /path/to/config.yaml for a custom location.

Upgrading from 0.4.4 preserves your endpoint, model, credentials, and calibration. No reinitialization, recalibration, or UR native runtime rebuild is required. To change models, open Settings → Model & runtime, connect, select, and save. Changes to the physical assembly, Home, or task axes still require recalibration.

Upgrade from 0.4.3 or earlierCopy Anchor Link

After updating the package above, complete the component and assembly settings introduced in 0.4.4:

1. Complete the saved installation configuration:

roboapi init --complete

For a custom location, use roboapi init --complete -c /path/to/config.yaml. This preserves existing settings and credentials and prompts for missing information.

2. Restart with the same configuration. Check video and device state, complete any requested calibration, then select Enable control. Use roboapi http for the default location, or roboapi http -c /path/to/config.yaml for another path.

Keep an existing UR native runtime from 0.4.2 or 0.4.3 in the same environment; no rebuild is required. A first-time UR e-Series installation in a new environment still builds the native runtime. Updating the Python package does not delete configuration, credentials, calibration, or run records. Recalibrate after changing the physical assembly, Home, or task axes.

Upgrade from 0.4.1 or earlierCopy Anchor Link

In addition to the steps above, rebuild the UR5e / UR7e native control runtime:

python -m roboapi.hardware.ur.build --build-dir build/ur-native --install-dir "$VIRTUAL_ENV"

For custom integrations, review the HTTP API and MCP tools. The browser connects Capture C, return Home, save, and enable control automatically; direct integrations perform the documented steps separately.

Migrate from the old robomcp nameCopy Anchor Link

1. Stop old tasks and services so two processes do not connect to the same device.
2. Create a new Python environment and install RoboAPI 0.4.5 using the installation guide.
3. Run roboapi init and enter component, device, load, and model settings.
4. For UR e-Series, build the new roboapi-ur-loop. An executable using the old name cannot replace it.
5. Complete Home and A/B/C calibration, then run a task.

RoboAPI uses ~/.config/roboapi/. It does not discover or migrate ~/.config/robomcp/; old files remain in their directory. Old commands and Python imports are not aliases for the new name.