Enterprise Angular 21 agent workspace prototype with a NestJS orchestrator, admin-governed tools, RAG-ready services, MCP-style integrations, SSE streaming hooks, Playwright/browser workers, and internal pilot operations.
This repo is a prototype and architecture proof project, not a claim of production enterprise adoption. It is designed to show how an internal AI agent workspace could be structured across frontend, backend, admin governance, and pilot rollout concerns.
- Angular 21 workspace shell with routes for agents, admin, pilot, ops, logs, readiness, and debug tools
- NestJS server modules for auth, agents, tools, RAG, connectors, pilot ops, observability, and browser workers
- role and permission guards for the frontend and backend
- Docker and Docker Compose setup for a local or pilot-style environment
- Prisma schemas for SQLite and PostgreSQL-oriented deployment paths
- mock-safe workspace behavior for agent sessions, artifacts, tool events, approvals, and test-worker style flows
- many agent responses and runtime events in frontend demo mode
- demo tools, demo workflows, demo approvals, and pilot feedback seed data
- browser and Playwright worker experience when used as a safe prototype path
- some admin and pilot dashboard content when running without a fully wired backend stack
- deeper production-grade provider integration and hardening
- fuller RAG citation and retrieval UX
- richer evaluation dashboards and agent scorecards
- multi-tenant controls and broader connector coverage
- stronger production observability and automated validation coverage
flowchart LR
User["Internal user"] --> Frontend["Angular 21 workspace"]
Frontend --> Agents["Agent workspace routes"]
Frontend --> Admin["Admin console"]
Frontend --> Pilot["Pilot hub"]
Frontend --> Ops["Ops and readiness views"]
Frontend --> Api["NestJS orchestrator API"]
Api --> Auth["Auth and RBAC"]
Api --> Orchestrator["Agent orchestration"]
Api --> RAG["RAG and context layer"]
Api --> Tools["Tool registry and approvals"]
Api --> MCP["MCP-style integrations"]
Api --> Browser["Browser and Playwright workers"]
Api --> Observability["Metrics, readiness, logs"]
Api --> Data["Prisma with SQLite or PostgreSQL"]
- Copy the environment template:
cp .env.docker.example .env-
Set required values in
.env, especially secrets such asPOSTGRES_PASSWORD, JWT values, and provider keys when needed. -
Start the stack:
docker compose up --build- Open the frontend:
http://localhost:8080
Use local development instead if you want to run frontend and backend separately.
Frontend:
npm install
npm startBackend:
cd server
npm install
npm run build
npm run start:dev/login/dashboard/agents/agents/:slug/admin/pilot/pilot/metrics/pilot/readiness/agent-readiness/ops/mcp-debug/internal-tools-debug/browser-test-debug
- agent workspace with chat, artifacts, approvals, tool window, and runtime events
- admin overview with visible mock tools, workflows, approvals, and platform summary
- pilot hub with rollout guidance, feedback loops, and readiness framing
- readiness report showing priority-agent status
Available assets:
docs/assets/screenshots/workspace.pngdocs/assets/screenshots/admin-dashboard.pngdocs/assets/screenshots/pilot-hub.pngdocs/assets/screenshots/readiness-report.pngdocs/assets/screenshots/tool-execution-timeline.pngdocs/assets/screenshots/org-ai-force-demo.gif
Capture guidance:
- no secrets should be committed to the repo
- admin and debug routes are permission-gated
- tool execution and approvals are treated as governed flows
- mock mode exists for safe demoing when live systems are unavailable
- Docker and environment templates are examples, not a substitute for real vaulting, HTTPS, or production secret handling
Start with:
- System architecture
- Agent orchestration
- RAG and tool layer
- Admin console
- Pilot rollout
- Security model
- Observability
- Recruiter review guide
For the fastest walkthrough, use:
Frontend:
npm run build
npm run typecheckBackend:
cd server
npm run build
npm run typecheckDocker:
docker compose config



