Specialist reviewer agents, named human gates, memory feedback loop, cost tracking, telemetry, audit-evidence packaging. Every box on the architecture diagram is already wired. Open the architecture page if you want the full diagram. Below is the five-step path most teams follow on day one.
npx great-cto init — auto-detects your stack, picks an archetype (ai-system / fintech / robotics / commerce / 26 more), generates a PROJECT.md.
If your repo touches voice / clinical / drug discovery / lending / HR-AI / climate / robotics / em-fintech / api-platform — the matching pack snaps on. Specialist reviewer agents enter the pipeline.
The pack reads its own template (200 words, vetted by three auditors), the regulation text, and your PROJECT.md. Drafts a first-pass threat model. You review in three hours instead of authoring it in three weeks.
gate:plan, gate:ship, gate:api-contract, gate:bias-audit, gate:hara-signoff. You approve. The pipeline records who, when, why. Auditors love this.
Per-project + per-org memory persists across sessions. Specialist agents stop re-litigating the same trade-off. Lessons-learned promoted to global decisions after ≥3 occurrences.
Compliance pack drops, reviewer-agent prompts, anonymized real-world cost numbers. Unsubscribe in one click. No course-selling.
$ npx great-cto init