This Go-based project integrates ChatGPT with DingTalk, enabling users to leverage AI for enhanced work efficiency within their DingTalk environment. It supports private and group chats, context-aware conversations, role-playing, and image generation, targeting professionals and teams seeking to automate tasks and improve communication.

How It Works

The bot operates via DingTalk's "stream" mode, processing messages received from DingTalk and forwarding them to OpenAI's API for processing. It supports various OpenAI models (including GPT-3.5, GPT-4, and GPT-4o-mini) and Azure OpenAI. The architecture allows for context management, role-playing via built-in prompts, and image generation, with configurable options for rate limiting, user whitelisting, and sensitive word filtering.

Quick Start & Requirements

• Installation: Docker (recommended), binary download, or local Go development.
• Prerequisites: OpenAI account with API key, DingTalk developer account to create an application and bot.
• Configuration: Requires setting environment variables or a config.yml file for API keys, model selection, DingTalk app credentials, and network proxy settings.
• Resources: Docker image available on Docker Hub. Local development requires Go.
• Documentation: Usage Guide, Configuration.

Highlighted Details

• Supports both OpenAI and Azure OpenAI endpoints.
• Offers "single chat" (stateless) and "continuous chat" (stateful) modes.
• Includes image generation capabilities using DALL-E 2 and DALL-E 3.
• Features role-playing via built-in prompt templates (e.g., for weekly reports).

Maintenance & Community

The project is actively maintained by eryajf and has seen contributions from a list of community members. A DingTalk group is available for discussion.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive license allows for commercial use and integration with closed-source applications.

Limitations & Caveats

The project documentation states that support for installation and configuration issues is limited, with paid technical support available. GPT-4's balance query and image generation features may have limitations due to API restrictions. The project primarily focuses on DingTalk's "stream" mode, with less emphasis on older HTTP modes.