A modular AI orchestration system supporting local LLM (llama.cpp) and API models, featuring dynamic routing, multi-stage processing pipelines, and persistent conversational memory.
Based on repository activity, growth velocity and community engagement.
35
Growth2/30
Activity10/25
Popularity7/25
Trust16/20
56
Stars
High
Sentiment
Votes
56
README.md
Xiaoxing AI
Xiaoxing AI is a production-style AI SaaS application for Gmail-driven workflows.
It combines a public marketing homepage, an authenticated user dashboard, an admin dashboard, Gmail and Telegram integrations, and a tracked AI usage and pricing system.
Overview
The application has three distinct product surfaces:
Public Homepage: public-facing landing page at /
User Dashboard: authenticated personal workspace at /dashboard for normal users
Admin Dashboard: authenticated analytics and operations dashboard at /dashboard for admins
Core capabilities:
Gmail processing pipeline with per-user OAuth and worker execution
Telegram bot integration for notifications and interactive flows
AI usage analytics persisted in ai_usage_analytics
Configurable pricing via AI_PRICING_JSON, API endpoints, and admin settings UI
Role-aware navigation, routing, and access control
Domain-oriented backend structure with domains/, core/, workflows/, and repository-based database access
Key Features
Public product landing page that remains separate from authenticated app navigation
Personal user dashboard for email activity, worker state, token usage, and estimated cost
Admin dashboard for global metrics, top users, model usage, and cost breakdown
AI usage tracking with per-request model, token, and estimated-cost analytics
Configurable pricing table with safe fallback defaults and admin editing support
Gmail worker system with user-scoped polling, processed email storage, and automation rules
Telegram bot management and delivery flows
Admin user management, invite-based registration, prompt management, and debugging tools
Architecture Summary
Backend:
app/api/routes/: HTTP route modules
: configuration, auth, app setup, realtime, and tooling infrastructure
app/core/
app/domains/: domain logic for Gmail, worker runtime, Telegram, and outgoing flows
app/workflows/: cross-domain orchestration such as the email processing flow
app/db/repositories/: SQL access layer grouped by resource
Authenticated user
→ /dashboard
→ role check from current user identity
→ admin
→ Admin Dashboard
→ global analytics, top users, cost breakdown, worker health
→ user
→ User Dashboard
→ personal email activity, token usage, estimated cost, worker status
Operational request flow
→ User action or worker poll
→ FastAPI route layer
→ domain logic / workflow orchestration
→ AI provider call
→ usage metadata returned
→ pricing lookup from centralized pricing source
→ analytics row written to ai_usage_analytics
→ persisted email/log/runtime data updated
Analytics outputs
→ User Dashboard reads personal aggregates from persisted user-scoped data
→ Admin Dashboard reads global aggregates from persisted system-wide data
Role separation
→ / stays public
→ /dashboard is authenticated
→ normal users never receive admin/global analytics
→ admins keep the global dashboard view
The docs/ tree is the maintained documentation source of truth. Legacy documentation folders have been removed to avoid duplication and stale architecture notes.
Ecosystem Role
Standard MoltPulse indexed agent.
Embed Badge
Show off your Pulse Score in your GitHub README to build trust and rank higher.