How code/app-nous works
Project type and purpose
• This is a full-stack TypeScript monorepo (“TypedAI”) providing:
• Backend API server (Fastify) exposing REST/HTTP routes for agents, chatbots, code reviews, workflows, authentication, etc.
• Frontend Angular web app serving UI for agent management, chat, code review, user profile.
• CLI tool for local agent/workflow invocation.
• Autonomous AI agent engine and LLM orchestration platform.
• Pluggable “functions” and integrations (filesystem, Slack, Jira, GitHub, Google Cloud, etc.).
• Persisted state via in-memory or Firestore, optional PostgreSQL for function caching.
• Observability via OpenTelemetry tracing.
File structure
• Top-level:
• frontend/ – Angular UI
• src/fastify/ – Fastify application, middleware, routes
• src/cli/ – CLI entrypoints and commands
• src/agent/, src/llm/, src/chat/, src/functions/, src/modules/, src/routes/, src/swe/, etc. – Core engine libraries, services, workflows, modules
• shared/ – shared config and route definitions
• docs/ – MkDocs site and CLI scripts
• Dockerfile, docker-compose.yml – container build & local dev
• .github/workflows – CI & integration tests
README for additional insights
• Confirms full-featured AI platform:
• Autonomous agents & software engineer agents
• Code review agents
• Chat interface + Slack chatbot
• CLI + Web UI
• Multiple LLM services supported (OpenAI, Anthropic, Gemini, etc.)
• Function schema generation from decorated methods
Be very detailed
• Mock-api module used by the UI in development.
• DevOps build/test pipelines (CI workflows, Docker images).
• Authentication middleware (JWT, SSO) and observability hooks.