What problem does it solve?
Google Takeout exports of Gemini activity arrive as messy ZIP archives with Unicode filenames, mixed media, and hundreds of transcripts that are impractical to review manually. This Skill turns that raw export into a structured, searchable analysis.
Core Features & Use Cases
- Unicode-safe extraction and inventory: Uses
unar to avoid macOS unzip corrupting Chinese filenames, then builds a full file-type and size inventory.
- Topic classification and conversation typing: Distinguishes meeting transcripts from prompt-response chats, classifies topics, and extracts key findings, decisions, and quotes per file.
- Domain keyword search with context verification: Greps for finance/legal/etc. terms, then reads surrounding context to eliminate false positives before reporting.
- Structured reporting and memory generation: Produces a Markdown analysis report with PII flagging, and can optionally write user-profile or project memory files.
- Use Case: Drag a Google Takeout ZIP into your project and ask what you discussed with Gemini about investing; the Skill extracts the archive, searches with context verification, and delivers a categorized report.
Quick Start
Analyze my Gemini Takeout export at ~/Downloads/takeout.zip and tell me what topics I discussed, with a focus on any finance-related conversations.