ikevss/wechat-ai-memory

What it solves

Many personal and professional contexts—such as project decisions, requirement changes, and key commitments—are scattered across WeChat private and group chats. This tool extracts these conversations into a structured format that can be used as a "memory bank" for AI agents, allowing LLMs to understand the full history and context of a project or event without relying on incomplete manual descriptions.

How it works

The tool reads a read-only temporary copy of the Windows WeChat 4.x database. It allows users to filter messages by contact, group, date range, and keywords. It then processes various media types: restoring original images, correcting EXIF orientation, and using a local faster-whisper model to transcribe SILK audio messages into text. Finally, it exports the filtered data into PDF, Markdown, or structured JSON formats for easy ingestion by AI models.

Who it’s for

Users of Windows WeChat 4.x who want to archive their chat history for personal knowledge management or to provide high-quality, structured context to AI agents and RAG workflows.

Highlights

• Local-First Privacy: All data processing, including database decryption and audio transcription, happens locally on the user's machine; no chat content is uploaded to any remote server.
• Multimodal Support: Handles text, images, files, and voice messages (with local transcription).
• AI-Ready Exports: Generates structured JSON and Markdown files specifically designed to be used as context for LLMs.
• Standalone Windows App: Provided as a portable executable that includes all necessary dependencies (Python, Qt, Frida) for non-technical users.