1. First Kernel¶

This walkthrough boots the smallest useful AAAX kernel.

1. Write a config¶

Create aaax.toml:

[aaax]
id = "aaax-main"
name = "AAAX Kernel"
policy = "default"

[aaax.libos]
name = "lllm"
strict_boot = true
discover_shared_packages = false

[aaax.network]
publish = false
host = "0.0.0.0"
port = 8100

No modules are required for the first boot.

2. Bootstrap from Python¶

import asyncio

from aaax import AAAXConfig, bootstrap_kernel


async def main() -> None:
    config = AAAXConfig.from_file("aaax.toml")
    kernel = await bootstrap_kernel(config, start_channels=True)
    print([channel.id for channel in kernel.all_channels])


asyncio.run(main())

You should see the kernel governance channels, including:

aaax.capability-request
aaax.action-gate
aaax.kernel-replies
aaax.module-loader
aaax.lifecycle
aaax.heartbeat

3. Step the kernel once¶

import asyncio

from aaax import AAAXConfig, bootstrap_kernel


async def main() -> None:
    config = AAAXConfig.from_file("aaax.toml")
    kernel = await bootstrap_kernel(config, start_channels=True)
    await kernel.step()


asyncio.run(main())

This processes any pending governance work and emits a heartbeat if that channel is active.

4. Use the CLI¶

aaax launch aaax.toml

That starts the kernel run loop. Use the Python path when you want embedding and programmatic control; use the CLI when you want a standalone kernel process.