Getting Started
Initial deployment, runtime binaries, workspace onboarding, and NVIDIA NIM setup.
"The model proposes. The runtime decides."Rust-native autonomous software engineering runtime with non-bypassable policy gates, verifiable execution, and isolated git attribution.
Organized reference manuals and engineering guides for operators, developers, and researchers.
Initial deployment, runtime binaries, workspace onboarding, and NVIDIA NIM setup.
Operator CLI workflows, interactive Ratatui cockpit, session persistence, and change reviews.
10-layer L0-L9 downward dependency discipline, runtime authority, and persistence models.
Rust development environment, crate hierarchy, module specifications, and test validation.
Process sandboxing with Landlock/seccomp, compile channels, diagnostics, and deployment.
Foundational architectural invariants and operational guarantees governing the M31A runtime.
The model proposes; the runtime decides. The LLM never writes to disk or executes shell commands directly without policy verification.
Inspect Authority InvariantsSingle canonical execution unit under task DAG scheduling, two-phase checkpointing, and complete state persistence across crashes.
View Execution LifecycleDedicated enterprise production inference wired exclusively to NVIDIA NIM (integrate.api.nvidia.com/v1) with token compaction.
Configure Model Provider7-tier verification hierarchy blocking task completion on model self-report, requiring cryptographic artifact snapshot matching.
Read Verification Manual