dropbox-kb-auto - SKILL.md Agent Skill

name: Dropbox KB Auto description: Your AI agent can't search your Dropbox. Documents, receipts, research papers, and notes sit in folders your agent doesn't know about — so you end up searching manually.

Dropbox KB Auto

What problem does this solve?

Your AI agent can't search your Dropbox. Documents, receipts, research papers, and notes sit in folders your agent doesn't know about — so you end up searching manually.

This skill bridges that gap: it syncs files from Dropbox, extracts text (PDF, Office, OCR for scans), and indexes everything into your agent's knowledge base. Changed files are detected via content hashing and automatically re-indexed. Unchanged files are never re-processed.

One-command installer configures folders, exclusions, file types, cron scheduling, and OpenClaw memory integration.

What This Skill Does


    ┌───────────────────┐

    │  Dropbox Account  │

    └────────┬──────────┘

             │ Delta API (only changes)

             ▼

    ┌───────────────────┐

    │  Text Extraction  │

    │  PDF, Office, OCR │

    └────────┬──────────┘

             │ Markdown files

             ▼

    ┌───────────────────┐

    │  OpenClaw Memory  │

    │  Embed + Search   │

    └────────┬──────────┘

             │

             ▼

    ┌───────────────────┐

    │  Agent answers    │

    │  your questions   │

    └───────────────────┘

Supported formats: PDF (with OCR fallback), Word, Excel, PowerPoint, images (Tesseract OCR), plain text.

Prerequisites

System Dependencies


# Debian/Ubuntu

sudo apt-get install -y tesseract-ocr tesseract-ocr-eng poppler-utils



# macOS

brew install tesseract poppler

Python Dependencies


pip3 install pypdf openpyxl python-pptx python-docx

Or run: bash setup.sh

Dropbox App Setup

Create app at https://www.dropbox.com/developers/apps
Choose Scoped access → Full Dropbox
Enable: files.metadata.read, files.content.read
Generate refresh token via OAuth 2 flow

Add to ~/.openclaw/.env:


DROPBOX_FULL_APP_KEY=your_app_key

DROPBOX_FULL_APP_SECRET=your_app_secret

DROPBOX_FULL_REFRESH_TOKEN=your_refresh_token

Installation

Via ClawHub


clawhub install dropbox-kb-auto

Manual


cd ~/.openclaw/workspace/skills

git clone https://github.com/ferreirafabio/dropbox-kb-auto.git

Configuration

Edit config.json:


{

  "folders": ["/Documents", "/Work"],

  "skip_paths": ["/Archive", "/Backups"],

  "file_types": ["pdf", "docx", "xlsx", "pptx", "jpg", "png", "txt"],

  "max_file_size_mb": 20

}

Usage

Interactive Setup (recommended)


./install.sh

Manual Sync


python3 dropbox-sync.py

First run: 5-10 min. Incremental runs: <10 sec.

Automated via Cron


openclaw cron create \

  --name "Dropbox KB Sync" \

  --cron "0 */6 * * *" \

  --timeout-seconds 14400 \

  --session isolated \

  --message "cd ~/.openclaw/workspace/skills/dropbox-kb-auto && python3 dropbox-sync.py"

Query Examples

Once indexed, ask your agent:

"Find presentations about machine learning"
"Search for expense receipts from Q1 2025"
"What do my project notes say about deployment?"

Performance

Tested on 650K files (1,840 indexable): first sync ~15 min, incremental <30 sec.

Security

Read-only Dropbox access recommended
All extraction runs locally
Credentials stored in ~/.openclaw/.env (gitignored)

License

MIT

Author

Fabio Ferreira (@ferreirafabio)