Summary

Murmur addresses the need for understanding personal communication data by transforming WeChat and QQ chat histories into interactive 3D relationship networks. It offers privacy-focused, local analysis and AI-driven insights for users seeking to visualize and interpret their social connections.

How It Works

The project decrypts WeChat and QQ data entirely locally, building a navigable 3D graph where contacts are positioned based on interaction frequency and duration. It integrates with local AI command-line interfaces (Claude/Codex) to generate detailed, evidence-backed relationship reports. Novel features include an "annual overview" akin to Spotify Wrapped and a unique dual-person relationship analysis.

Quick Start & Requirements

• Install: Downloadable executables for Windows (.exe, .msi) and macOS (.dmg, .app.zip). Source build required for Intel Macs.
• Prerequisites:
  • Windows: Specific WeChat versions (not App Store). QQ NT decryption supported. Antivirus whitelisting may be needed for wx_key.dll.
  • macOS: Official Tencent WeChat for Mac (not App Store version). May require Gatekeeper bypass via Terminal command (xattr -dr com.apple.quarantine).
  • AI: Local Claude/Codex CLI must be installed and logged in.
• Setup: Initial setup involves a one-time key-grabbing process requiring WeChat/QQ login. Decryption can be time-consuming for large histories, but incremental caching speeds up subsequent refreshes.
• Links: Latest recommended version: v0.4.3. Video tutorials and detailed installation guides are available within the README.

Highlighted Details

• Annual Overview: Generates a "Spotify Wrapped"-style retrospective including top contacts, monthly highlights, busiest days, chat heatmaps, and relationship dynamics.
• AI Relationship Reports: Leverages local AI models to produce in-depth analyses of individual friendships or pairs, using anonymized chat samples and derived data fingerprints.
• Dual-Person Relationship Archive: Analyzes interactions between any two contacts (shared groups, private messages, Moments) and exports findings in MD, JSON, or HTML formats.
• Privacy Mode: Toggles visibility of real names, IDs, and sensitive text patterns for secure sharing or screenshots.
• WeChat/QQ Enhancements: Includes features like WAL merging, incremental decryption caching, atomic swaps, and specific support for Windows QQ NT.

Maintenance & Community

No specific details regarding maintainers, community channels (e.g., Discord, Slack), or roadmap are provided in the README. Support is directed to GitHub issues.

Licensing & Compatibility

The project's license is not explicitly stated in the README. Local analysis ensures data privacy, but AI report generation relies on external CLIs with their own terms. Compatibility is restricted to specific versions of WeChat and QQ.

Limitations & Caveats

The macOS App Store version of WeChat is unsupported. Intel Macs require source compilation. Antivirus software may interfere with the key decryption process. The effectiveness of AI reports depends on the user's local AI setup.