AI Email Agent#

LLM 驱动的电商客服智能邮件自动回复系统。自动分类、情感检测、智能回复、紧急升级、多语言支持。

Quick Start#

bash

# Install dependencies
pip install -r {baseDir}/requirements.txt

# Demo mode — process 5 sample emails
python3 {baseDir}/main.py demo

# Run agent loop (requires IMAP/SMTP config)
python3 {baseDir}/main.py run

# Start monitoring dashboard
python3 {baseDir}/main.py dashboard

Configuration#

Edit {baseDir}/config.yaml and set environment variables:

bash

export LLM_API_KEY="sk-your-key"
export IMAP_PASSWORD="your-imap-password"
export SMTP_PASSWORD="your-smtp-password"

Features#

ModuleDescriptionSmart ClassificationLLM-powered 5-category routingSentiment Detection5-level sentiment + urgency scoringAuto ReplyMulti-template + variable renderingEscalation10 trigger rules + WeCom/Feishu notificationsMulti-languageP0-P3 tiered strategy for 8+ languagesKnowledge BaseKeyword RAG for consultation enhancementDashboardReal-time Web monitoring with Chart.js

Decision Matrix#

text

               Urgency →
Sentiment↓     1(Low)  2(Normal) 3(Medium) 4(High)  5(Critical)
S0 Positive    Auto    Auto      Auto      Auto+CC   Auto+CC
S1 Neutral     Auto    Auto      Auto      Semi      Semi
S2 Negative    Auto+   Auto+     Semi      Human     Human
S3 Angry       Semi    Semi      Human     Human     Human+Legal

Commands#

CommandDescriptionpython3 {baseDir}/main.py demoDemo with 5 sample emailspython3 {baseDir}/main.py runContinuous agent looppython3 {baseDir}/main.py onceSingle fetch & processpython3 {baseDir}/main.py dashboardWeb dashboard (:8080)python3 {baseDir}/main.py statsTicket statisticspython3 {baseDir}/main.py test-send <email>Test SMTP config