wangrongding/wechat-bot

This system is a command-line interface for a WeChat bot that aggregates multiple AI providers and messaging channels. It enables users to interact with various large language models, such as OpenAI, Claude, and local Ollama instances, through unified adapters for platforms like WeChat, Telegram, and Lark. The application also includes utilities for analyzing local WeChat chat logs and routing messages to different AI services based on user configuration.

5 changes.

The project was initialized with core infrastructure, including CLI tools, Docker configurations, and dependency management. Subsequent work focused on integrating a wide array of AI providers, such as OpenAI, ChatGPT, and various Chinese models, into a unified WeChaty bot architecture with dynamic service routing.

13 changes.

The project expanded its capabilities by integrating a wide array of AI models, including Kimi, iFlytek, DeepSeek, Doubao, 302.ai, Dify, Ollama, Tongyi, and Claude, alongside new messaging channels like Lark, Telegram, and WhatsApp. A unified CLI entry point was introduced to manage these diverse services and channels, while supporting features such as streaming responses and multimodal inputs. The period also included infrastructure improvements like automated code formatting and utilities for WeChat chat analysis.

Introduces a new module in src/302ai that connects to the 302.ai API to generate automated responses. The implementation uses the GPT-4o-mini model by default, configurable via the \_302AI\_MODEL environment variable, and requires the \_302AI\_API\_KEY for authentication. A test script is included to verify the integration.

src/302ai · high confidence

Introduces a new module for interacting with the Claude API, allowing users to configure the service endpoint via the CLAUDE\_BASE\_URL environment variable. The implementation includes logic to normalize the base URL to the standard Anthropic format, supports custom models (defaulting to claude-3-5-sonnet-latest), and enables system prompt configuration through CLAUDE\_SYSTEM.

src/claude · high confidence

Introduces a new module in src/dify that enables the application to send chat messages to a Dify AI service. The implementation uses environment variables (DIFY\_API\_KEY, DIFY\_URL, BOT\_NAME) to configure the API endpoint and authentication, exposing a getDifyReply function that sends a user prompt and returns the assistant's response in blocking mode.

src/dify · high confidence

Users can now interact with the Doubao language model by configuring environment variables (DOUBAO\_API\_KEY, DOUBAO\_URL, DOUBAO\_MODEL) and using the new getDoubaoReply function. The implementation leverages the OpenAI client library to support both text-only prompts and multimodal inputs that include image URLs, allowing users to send queries with or without attached images.

src/doubao · high confidence

Users can now connect to an Ollama instance via the new src/ollama module. The integration sends requests to the configured OLLAMA\_URL using the specified OLLAMA\_MODEL and includes a configurable OLLAMA\_SYSTEM\_MESSAGE as a system role in the conversation history. A test script is provided to verify the connection and response handling.

src/ollama · high confidence

A new module at src/tongyi/index.js has been added to enable interaction with the Tongyi (Qwen) language model. This component initializes an OpenAI-compatible client using environment variables (TONGYI\_URL, TONGYI\_API\_KEY, and TONGYI\_MODEL, defaulting to 'qwen-plus'), validates the presence of a .env file, and exposes a getTongyiReply function that sends user prompts to the model and returns the response, automatically appending a request to answer in Chinese.

src/tongyi · high confidence

This change introduces new modules for analyzing local WeChat chat logs and integrating with the Pi agent via IM channels. The \src/analysis\ directory now contains \wechatAnalyzer.js\, which provides functions to load, filter, and generate statistics (such as top speakers and message counts) from WeChat message records, along with a test suite in \\_\test\\_.js\ to verify this logic. It also constructs prompts for an AI service to generate textual analysis of the chat data. Additionally, \src/pi/index.js\ implements the \getPiReply\ function, which acts as the entry point for the Pi agent to handle user queries through the IM channel, instructing the agent to leverage local WeChat analysis capabilities when appropriate. Utility functions for running and streaming shell commands have been added in \src/utils/process.js\ to support these operations.

src/analysis, src/pi, src/utils · high confidence

Users can now interact with DeepSeek and Doubao models using the OpenAI API format. The update introduces new modules in src/deepseek and src/deepseek-free that handle API communication, including configuration for custom URLs, API keys, and system prompts via environment variables. This enables the application to leverage these specific AI providers through a standardized interface.

src/deepseek-free · high confidence

The repository has been initialized with the core project structure, including a CLI entry point (\cli.js\), Docker build files (\Dockerfile\, \Dockerfile.alpine\), and a comprehensive environment configuration template (\.env.example\) that defines settings for multiple AI providers (OpenAI, Doubao, DeepSeek, Kimi, Xunfei, Dify, Ollama, Claude, Pi) and IM channels (WeChat, Lark, Telegram, WhatsApp). Documentation (README, LICENSE) and development tooling (Prettier, .npmrc) have also been added to support the project's setup.

(repo-wide) · high confidence

Users can now interact with the iFlytek Spark large language model via a new integration in the \src/xunfei\ module. The system connects to the iFlytek API using WebSocket, allowing configuration of the model version (defaulting to v4.0) and a custom system prompt through environment variables. This change introduces the capability to send messages and receive AI-generated replies from the iFlytek service.

src/xunfei · high confidence

Added a new module at src/chatgpt/index.js that initializes the ChatGPT API client using environment variables for authentication (clearance token, session token, user agent, access token) and provides a function to send messages with a 2-minute timeout.

src/chatgpt · high confidence

Added a new \src/kimi\ module that integrates with the Moonshot API to generate replies. The implementation enables streaming output by default, allowing users to receive incremental responses, and sets a default maximum token limit of 5000 for generated content. A test file is included to verify the \getKimiReply\ function.

src/kimi · high confidence

The src/openai module now provides a direct integration with the official OpenAI API, replacing previous third-party wrappers. Users can configure the API key, an optional proxy URL via OPENAI\_PROXY\_URL, and a system message. The model is configurable via the OPENAI\_MODEL environment variable (defaulting to gpt-4o), and responses are automatically stripped of Markdown formatting before being returned.

src/openai · high confidence

The WeChaty bot now supports a wider variety of AI providers through a new service selection mechanism. The \serve.js\ module introduces a \getServe\ function that dynamically loads responses from multiple providers including ChatGPT, Kimi, Xunfei, DeepSeek, 302AI, Dify, Ollama, Tongyi, Claude, Pi, and Doubao. The \sendMessage.js\ file implements new logic for handling incoming messages, distinguishing between group chats and private chats, and applying prefix-based auto-reply rules and command handling. A new \testMessage.js\ utility allows users to manually test these different AI services via an interactive CLI prompt.

src/wechaty · high confidence

The application now exposes a new command-line interface via src/index.js, allowing users to start various AI services (including ChatGPT, Doubao, DeepSeek, Kimi, Xunfei, 302AI, Dify, Ollama, Tongyi, Claude, and Pi) and IM channels (WeChat, Lark, Telegram, and WhatsApp) through a unified CLI. The entry point handles environment variable validation for each service and channel, providing specific error messages for missing configurations, and supports both interactive selection and direct service specification via command-line options.

src · high confidence

The application now integrates with Lark (Feishu), Telegram, and WhatsApp, allowing users to interact with the bot via these platforms. This update adds specific adapters for each service (including Lark CLI integration, Telegram long-polling, and WhatsApp webhooks) and platform agents that handle message routing, deduplication, and reply logic. Configuration for these channels is centralized in the environment config, requiring new environment variables such as TELEGRAM\_BOT\_TOKEN, WHATSAPP\_ACCESS\_TOKEN, and LARK\_CLI\_BIN to enable the respective features.

src/adapters, src/config, src/platforms · high confidence

A pre-commit hook has been added to the repository to automatically run lint-staged. This ensures that code is consistently formatted before changes are committed, reducing manual review overhead for style issues.

.husky · high confidence

The project has replaced the legacy yarn.lock with npm-style dependency management and significantly updated its package.json. Key changes include upgrading wechaty to v1.20.2, adding the official openai SDK (v4.52.0) and axios (v1.6.8), and introducing new dependencies for CLI interaction (commander, inquirer) and text processing (remark, strip-markdown). The package name was corrected to 'wechat-bot', version bumped to 1.0.2, and the module type set to 'module'. Development tooling was added with husky, lint-staged, and prettier for code formatting and pre-commit hooks.

(dependencies) · high confidence

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

• CAI 40 → 41 (+0.6)
• Rubric changed (rubric-2026.09.12 → rubric-2026.09.18) — scores are not directly comparable.

• Code Health 52 → 52 (+0.0)
• Architecture 95 → 95 (+0.8)
• Maturity 56 → 56 (+0.1)
• Readiness 15 → 15 (+0.0)
• Security 97 → 98 (+0.5)
• Performance 85 (new)

• Documentation: no installation or build instructions (README.md)
• Documentation: no usage examples (README.md)
• Hotspot: src/index.js (src/index.js)
• Off-boarding risk: anonymized user #1

• Documentation: no architecture or design documentation (README.md)
• Off-boarding risk: anonymized user #1

Written by watchdog.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

wangrongding/wechat-bot was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.