The Runtime, Addressed by Name.
M31A provides a complete clap-derive CLI with machine-readable --output json options, strict UNIX exit codes, and an interactive Ratatui TUI cockpit.
Interactive Command Preview
Select commands below to inspect real documented output and environmental checks.
Standardized UNIX Exit Codes
Every invocation returns a predictable exit code designed for headless CI/CD evaluation runners:
Mission or command finished with all verification checks satisfied.
Compiler error, failing tests, or unverified diffs.
Operation blocked by policy gate (fail-closed DENY).
Exceeded tokens, execution time, steps, or financial limits.
Process fault or panic; state preserved in SQLite.
Malformed config file or invalid authority parameter.
Full Command Catalog
18 documented subcommands across 6 operational categories.
Launch the interactive Ratatui cockpit interface.
Start an autonomous software engineering mission.
List all recorded missions in the current workspace.
Display detailed mission status, task DAG, and budget consumption.
Control the lifecycle state of an active mission.
Inspect individual tasks in the mission DAG and their verification evidence.
Inspect active and idle agents across the 8 canonical roles.
List all 15 core capability families and operational availability.
Dry-run evaluate an action or tool against active policy layers.
Manage atomic two-phase mission checkpoints and rollback.
Inspect content-addressed immutable execution artifacts and SHA-256 digests.
Run comprehensive environmental diagnostic probes.
Inspect execution traces, hierarchical spans, metrics, and token usage.
Inspect, validate, and manage hierarchical configuration across 7 tiers.
Execute autonomous evaluation harness benchmark scenarios.
Initialize workspace onboarding state (idempotent; use --force to re-enter setup).
Display deployment channel, build identity, commit hash, and artifact details.
Transactional binary update from manifest and atomic rollback to previous binary.