Bus Engine architecture

Bus Engine architecture

Bus Engine separates user-facing control, agent coordination, deterministic Linux operations, and evidence storage.

Bus Engine UI / CLI / external API
        ↓
Bus Engine orchestration, policy, and audit layer
        ↓
Agent runtime adapter
        ↓
Agent runtime
        ↓
Customer-selected hosted or local model provider

Bus Engine Linux tools
        ↓
Isolated build environment
        ↓
Boot and validation environment
        ↓
System blueprint and evidence

Bus Engine currently integrates Codex App Server as a separate open-source agent-runtime component. Bus Engine supplies Linux-specific tools, system blueprints, policies, approval rules, inventory and inspection tools, kernel configuration tools, package and image builders, isolated build environments, boot and test environments, validation tools, logs, and evidence.

The orchestration layer invokes the Linux tools and passes their outputs back into the blueprint and evidence store. The agent runtime uses that evidence to plan the next change, while deterministic build and validation steps produce the artifacts and test results that the orchestration layer records.

Codex App Server uses bidirectional JSON-RPC 2.0. Bus Engine may expose its own HTTP APIs to user interfaces and external systems, but those APIs are not the native App Server protocol.

Lifecycle

Bus Engine uses the same lifecycle across preview targets:

  1. Discover declared requirements and available evidence from target profiles, running Linux systems, hardware inventories, workloads, source repositories, build systems, configuration, logs, and acceptance tests.
  2. Infer required capabilities, dependencies, kernel features, packages, services, boot components, filesystems, policies, and validation requirements.
  3. Configure a versioned system blueprint covering the kernel, userspace, packages, services, boot process, filesystems, networking, security configuration, target profile, build outputs, and tests.
  4. Build kernels, packages, filesystems, and images with deterministic tools.
  5. Verify the generated system by booting it in an isolated environment and running platform and workload tests.
  6. Diagnose and repair failed builds, boot failures, missing dependencies, kernel problems, service failures, and test regressions.
  7. Maintain the blueprint as workloads, source code, kernels, packages, toolchains, policies, and upstream projects change.

AI performs the adaptable engineering. Deterministic tools perform and verify the repeatable operations.

Permissions and approvals

The agent may inspect, decide, modify, build, diagnose, and repair within the permissions and policies configured for the project. High-impact operations can require explicit approval, while lower-risk workflows may be automated according to customer policy.

Incomplete evidence creates uncertainty. Bus Engine reasons from customer intent, target inventories, source analysis, runtime observations, logs, tests, and previous build and maintenance evidence.

← Bus Engine overview Bus Engine Status and roadmap →