Tell us what you need. Our AI team will build and deliver it.
Codanium is a full-service AI software delivery platform. Users describe what they want built and a team of 24 specialized AI agents handles the entire development lifecycle — requirements, architecture, scaffolding, design, coding, testing, security review, and deployment.
The user is the stakeholder, not a developer. No technical background required.
| Resource | URL |
|---|---|
| Production App | https://codanium.com |
| GitHub (Web App) | https://github.com/AiSenseiMY/Ai-Team_studio |
| GitHub (Desktop) | https://github.com/AiSenseiMY/Codanium |
| Desktop Downloads | https://github.com/AiSenseiMY/Codanium/releases |
| API Health Check | https://codanium.com/api/llm/health |
This is a monorepo. Each top-level directory is an independent piece of the platform:
.
├── ai-team-studio/ # Web application (Next.js 16, Prisma, PostgreSQL, BullMQ)
├── codanium-desktop/ # Desktop app (Tauri v2, React 19) — primary client
├── ai-team-studio-vscode/ # VS Code extension (deprecated; replaced by codanium-desktop)
├── AI_Team_Studio_v4/ # Enterprise architecture documentation (20 docs)
└── README.md / LICENSE / CONTRIBUTING.md / CODE_OF_CONDUCT.md / SECURITY.md
cd ai-team-studio
# Start PostgreSQL + Redis (Docker required)
docker compose up -d
# Install dependencies
npm install
# Sync database schema
npx prisma db push
# Seed with sample data
npx prisma db seed
# Start dev server (http://localhost:3000)
npm run devcd ai-team-studio
docker compose up -d # Starts app (14001), worker, postgres (14000), redis (14003)| Layer | Technology |
|---|---|
| Framework | Next.js 16 (Turbopack, React 19, App Router) |
| Database | PostgreSQL via Docker (port 14000), Prisma 7.x |
| Auth | NextAuth.js (dev mode auto-login) |
| Styling | Tailwind CSS 4, dark theme, amber accent |
| State | Zustand, framer-motion |
| Components | Radix UI, Recharts, cmdk, @dnd-kit |
| AI Backend | Multi-provider LLM — OpenAI / Anthropic / Ollama / NVIDIA / Mistral / Groq |
| Encryption | AES-256-GCM (API keys at rest) |
| Task Queue | BullMQ + Redis for background agent work |
User describes idea (web app)
→ PM (Product Manager) activates as gatekeeper
→ PM creates card → BA collects requirements → BRD → PM validates
→ PM creates card → SA designs architecture → SDD → PM validates
→ PM creates card → DO scaffolds project → verifies build passes
→ PM hands to TL → UX creates UI Kit → UI Designer builds interfaces
→ User approves UI in "Designs" menu
→ PM confirms all gates passed (BRD + SDD + Scaffold + UI)
→ TL assigns ONE feature card → JD/SD codes → tsc validation
→ QA + SEC + DO + PE quad sign-off → card DONE
→ TL picks next card → repeat until all features built
→ PM validates → deployment
- PM is gatekeeper — creates all phase cards, validates all outputs
- One card at a time — dev finishes one feature fully before starting the next
- Quad sign-off — QA + SEC + DO + PE must all approve before a card is DONE
- Everything on the board — all work tracked as cards on the Work Board
- Code must compile — dev runs
tsc --noEmitbefore requesting sign-off
| Agent | Role |
|---|---|
| PM | Gatekeeper — creates all phase cards, validates outputs, controls pipeline flow |
| BA | Requirements gathering — discovers needs from user, generates BRD |
| SA | Architecture — designs system, generates SDD |
| DO | DevOps — project scaffolding, deployment, infrastructure sign-off |
| UX | UX Designer — brand identity, color palette, typography, UI Kit |
| UID | UI Designer — builds page layouts, wireframes, components using UI Kit |
| TL | Tech Lead — manages UX/UI and dev card assignments, coordinates pipeline |
| JD | Junior Developer — implements features, writes code |
| SD | Senior Developer — complex features, architecture-level code |
| QA | QA Engineer — code review, test strategy, functional sign-off |
| SEC | Security — vulnerability review, security sign-off |
| PE | Platform Engineer — infrastructure readiness, scaling sign-off |
| Agent | Role |
|---|---|
| ORC | Orchestrator — message routing, intent classification |
| STC | State Controller — enforces card lifecycle rules |
| DEC | Decision Controller — creates/manages decision records |
Configure multiple LLM providers with priority order in Admin Settings:
- Primary → Fallback 1 → Fallback 2 → Last Resort
- If primary fails (rate limit, timeout), automatically tries next provider
- Each provider stores: name, API key (encrypted), base URL, model, priority
OpenAI, Anthropic, Ollama, NVIDIA, Mistral, Groq, Together, Custom (OpenAI-compatible)
- User-level BYOK config (user's own key)
- Agent-level override (admin per-agent)
- Project-level override (admin per-project)
- Platform fallback chain (ordered by priority)
- Admin default settings (last resort)
| Phase | Status |
|---|---|
| Phase 1 — Foundation | Done |
| Phase 2 — Full SDLC Pipeline | Done |
| Phase 2.5 — Pipeline Hardening (PM gatekeeper, quad sign-off, scaffolding) | Done |
| Phase 3 — Enterprise (auth, billing, WebSocket, GitHub PR, deploy) | In Progress |
| Phase 4 — Custom Desktop App (replacing VS Code extension) | Planned |
| Phase 5 — Scale (marketplace, multi-tenant SaaS, white-label) | Planned |
- PM-gatekeeper SDLC pipeline with 6 phases (Requirements → Architecture → Scaffolding → UI/UX → Development → Release)
- 15 pipeline agents + 3 internal agents with system prompts and authority levels
- Multi-LLM provider fallback with encrypted API key storage
- Model-agnostic chat sanitization (strips tool call syntax from all LLM providers)
- Card lifecycle with quad sign-off (QA + SEC + DO + PE)
- Card comments + sign-off tracking (Jira-style audit trail)
- Token usage tracking per card
- Semantic card deduplication (70% word overlap detection)
- Loop detection with Decision record creation (My Decisions menu)
- "Designs" menu for UX/UI approval workflow
- Real code generation — agents write actual files to disk
- AES-256-GCM API key encryption
- Drag-and-drop kanban board, command palette, onboarding wizard
- Production authentication (currently NextAuth dev/demo mode)
- Stripe billing & subscriptions
- Custom desktop app (replacing VS Code extension)
- GitHub/GitLab PR creation
- Cloud deployment pipeline
Full enterprise architecture docs in AI_Team_Studio_v4/ — 20 documents covering vision, SDLC model, agent contracts, security, observability, architecture diagrams, and roadmap.
Test reports from SDLC pipeline testing in ai-team-studio/test-reports/.
We welcome contributions! See CONTRIBUTING.md for setup instructions per subproject and the PR workflow. By participating you agree to abide by the Code of Conduct.
Found a security issue? Please report it privately — see SECURITY.md. Do not open a public issue for vulnerabilities.
MIT © 2026 AiSenseiMY
Last updated: 2026-05-10