Run a self-verifying desktop AI agent on your own machine: a 10-minute quickstart

Description and use

OpenAmer is a local AI agent that acts on desktop windows and writes receipts before each action, enabling post-hoc verification of its behavior.

Problem

Deploy a local AI agent that can act on desktop windows while providing verifiable receipts of its actions.

Outcome

The agent can autonomously perform desktop tasks, logs each action with a pre-execution receipt, and provides a durable record that can be audited after the fact.

Workflow

• Install OpenAmer via shell or PowerShell script
• Run openamer setup to configure model provider and optional OAuth
• Use openamer asi think/learn/remember to invoke reasoning, learning, and memory subsystems
• Drive desktop by controlling a persistent Chrome profile or native windows via DevTools protocol
• Observe that the agent writes receipts to an outcome ledger before performing actions
• Verify receipts to confirm actions succeeded or failed
• Optionally run multiple instances to explore A2A Global Mesh features

Evidence

OpenAmer is an open-source (Apache-2.0), Windows-first personal agent.

Source

Every action it takes is logged with a pass/fail you can check afterwards.

Source

The agent writes receipts before the side effect, so an already-landed action becomes a no-op instead of a double action.

Source