Quick Start
Configure credentials, launch the cockpit, and run your first mission.
Get up and running with M31A in your terminal in under 5 minutes.
1. Configure NVIDIA NIM Credentials
Production Provider Invariant: M31A production supports NVIDIA NIM exclusively (
nvidia_nim). Retired provider IDs (openai, anthropic, gemini, ollama) are rejected deterministically.Set the environment variable or persist it locally with 0600 file permissions:
terminal
# Option A: Environment variable
export NVIDIA_API_KEY="nvapi-..."
# Option B: Persist in local user credentials
m31a config set-credential nvidia_nim "nvapi-..."2. Run Your First Autonomous Mission
Execute a targeted refactoring or coding task in your workspace:
terminal
# Run an autonomous coding mission under the "coding" profile
m31a mission run "Refactor database queries in src/repo to use parameterized statements" --profile coding3. Launch the Interactive Cockpit (TUI)
Launch the full Ratatui 0.30 terminal cockpit for interactive pairing, task DAG inspection, and governance reviews:
terminal
m31a tui4. Inspect Mission Telemetry & Resource Usage
Query the SQLite event index and token metrics for a completed mission:
terminal
m31a telemetry inspect <MISSION_ID> --summary --spans --metrics5. Environment Health Diagnostics
Verify platform prerequisites (Git worktree support, POSIX rlimits, cgroups v2, compiler):
terminal
m31a doctor