Dingtalk to ai skill

Dingtalk Bridge Skill

This skill manages a bridge service that forwards messages from Dingtalk group chats to local AI CLI tools (OpenCode or Claude Code), allowing remote AI interaction from mobile devices via Dingtalk.

Features

• Chat Commands: /help, /status, /model, /history, /queue, /config, /reset, /remember
• Project Memory: Auto-summarize conversations, inject context across sessions
• Media Processing: Voice transcription, image description (configurable)
• Multi-Agent Routing: Route messages to different AI providers by rules
• Scheduler: Cron-based scheduled message processing
• Interactive Cards: Dingtalk interactive card message support

Quick Reference

Command	Description
setup, 配置	Interactive configuration wizard
start, 启动	Start the bridge service
stop, 停止	Stop the bridge service
status, 状态	Check if service is running
logs [N], 查看日志	Show last N lines of logs (default: 50)
doctor, 诊断	Run diagnostics and troubleshoot issues
rebuild, 重新构建	Rebuild the TypeScript project

Workflow

1. Parse Command

Extract the subcommand from $ARGUMENTS based on user input (supports both English and Chinese keywords).

2. Configuration Check

For all commands except setup:

• Check if .env file exists in project root
• If missing: guide user to run setup first
• If present: proceed with command

3. Environment Detection

Detect which environment Claude is running in:

• Claude Code: Full features available including interactive prompts
• Other/Codex: Limited features, provide manual instructions instead

4. Execute Command

setup: Interactive wizard to configure Dingtalk credentials and AI provider preference

start: Build the project and start the bridge service via PM2

stop: Stop the PM2-managed bridge service

status: Report service health and connectivity

logs: Display service logs with optional line count

doctor: Run comprehensive diagnostics (env, deps, build, Dingtalk API connectivity)

rebuild: Compile TypeScript and verify build success

Examples

Interactive Configuration

User: "配置钉钉" AI: 执行 setup 命令，启动交互式配置向导

Start the Bridge

User: "启动钉钉桥接服务" / "start bridge" AI: 执行 start 命令，构建项目并通过 PM2 启动服务

Check Service Status

User: "钉钉桥接状态如何" AI: 执行 status 命令，报告服务健康状态

View Logs

User: "查看最近100行日志" AI: 执行 logs 100 命令

Run Diagnostics

User: "钉钉桥接有问题" AI: 执行 doctor 命令，进行全面诊断

Edge Cases

• Missing .env file: Guide user to run setup first
• PM2 service not found: Provide rebuild instructions with rebuild command
• Dingtalk API connection failed: Show specific error and suggest doctor command
• Service already running: Inform user and suggest stop first
• Build failed: Display error messages and suggest checking dependencies

Important Notes

• Always mask secrets in output (show only last 4 characters)
• The service defaults to Stream mode, falling back to Polling if needed
• Configuration persists in .env file with restricted permissions (600)
• Project memory is enabled by default, auto-summarizes after 20 messages
• Media processing (voice/image) is disabled by default, enable via MEDIA_VOICE_TRANSCRIPTION=true / MEDIA_IMAGE_DESCRIPTION=true
• Multi-agent routing is disabled by default, enable via ROUTER_ENABLED=true
• Scheduler is disabled by default, enable via SCHEDULER_ENABLED=true
• Project directory: $PROJECT_DIR (current working directory)