gog-wrapper

name	gog-wrapper
description	CLI wrappers for Google Workspace (Gmail, Calendar, Drive, Docs, Sheets, Photos) via gogcli. Use when the user asks you to read or write email, calendar events, Drive files, Docs, or Sheets.
metadata	{"openclaw":{"requires":{"bins":["gog","jq"]}}}

gog-wrapper

These bash scripts wrap gogcli (invoked as gog) to give you CLI access to Google Workspace. Use them whenever the user asks you to read or write Gmail, Google Calendar, Drive, Docs, Sheets, or Photos on their behalf.

All scripts live in this directory. Run them with bash <script>.sh or ./script.sh if executable. They all source vars.sh for the default account.

Install

# 1. Clone the repo
git clone [e-mail]:stoleas/gog-wrapper.git
cd gog-wrapper

# 2. Install gogcli
brew install steipete/tap/gogcli

# 3. Set your Google account
#    Edit vars.sh and set DEFAULT_GOG_ACCOUNT
vi vars.sh

# 4. Make scripts executable
chmod +x *.sh

# 5. Authorize gog with your Google account
gog auth credentials /path/to/client_secret.json
gog auth add [e-mail] --services gmail,calendar,drive,contacts,docs,sheets

# 6. (Optional) For analyze-for-auction.sh — pull a vision model on your Ollama server
ollama pull llava

analyze-for-auction.sh extra dependencies

# jq (JSON processor)
brew install jq           # macOS
sudo apt install jq       # Ubuntu/Debian

# python3 (for JSON extraction — usually pre-installed)
python3 --version

Install as a Claude Code skill

mkdir -p ~/.claude/skills
ln -s "$(pwd)" ~/.claude/skills/gog-wrapper

Claude will pick up SKILL.md automatically from ~/.claude/skills/gog-wrapper/.

Install as an OpenClaw skill

# Install OpenClaw (requires Node 24 or 22.14+)
npm install -g openclaw@latest
openclaw onboard --install-daemon

# Symlink this repo into the OpenClaw skills directory
mkdir -p ~/.openclaw/skills
ln -s "$(pwd)" ~/.openclaw/skills/gog-wrapper

OpenClaw hot-reloads skills — no restart needed after the symlink. To verify it loaded:

openclaw doctor

To enable the skill and pass required env vars, add to ~/.openclaw/openclaw.json:

{
  skills: {
    entries: {
      "gog-wrapper": {
        enabled: true,
        env: {
          GOG_ACCOUNT: "[e-mail]"
        }
      }
    }
  },
  agents: {
    defaults: {
      skills: ["gog-wrapper"]
    }
  }
}

Account

The active Google account is set by GOG_ACCOUNT or DEFAULT_GOG_ACCOUNT in vars.sh. Override per-command with -a <email>.

Output modes

Flag	When to use
--json	Scripting — structured output, pipe to jq
--plain	TSV — stable columns, pipe to awk/cut
--results-only	With --json — strips wrapper, emits array directly
--dry-run	Preview what would happen without executing
--no-input	CI/non-interactive — fail instead of prompting

gmail.sh

Wraps gog gmail. Use for reading and sending email.

# Search (Gmail query syntax)
./gmail.sh search 'is:unread newer_than:1d' --max 20 --json

# Read a specific message
./gmail.sh get <messageId> --json

# Send
./gmail.sh send --to [e-mail] --subject "Subject" --body "Body"
./gmail.sh send --to [e-mail] --subject "Subject" --body-file ./body.txt

# Organize
./gmail.sh archive <messageId>
./gmail.sh mark-read <messageId>
./gmail.sh trash <messageId>

# Batch archive old mail
./gmail.sh archive $(./gmail.sh search 'older_than:30d' --json --results-only | jq -r '.[].id')

# Drafts
./gmail.sh drafts create --to [e-mail] --subject "Draft" --body-file draft.txt
./gmail.sh drafts list --json

calendar.sh

Wraps gog calendar. Use for reading, creating, and managing calendar events.

# List calendars
./calendar.sh calendars --json

# List events in a date range
./calendar.sh events primary --from 2026-04-23T00:00:00Z --to 2026-04-30T00:00:00Z --json

# Get a single event
./calendar.sh event <calendarId> <eventId> --json

# Create an event
./calendar.sh create primary --summary "Standup" --from 2026-04-24T09:00:00Z --to 2026-04-24T09:30:00Z

# Update / delete
./calendar.sh update <calendarId> <eventId> --summary "New title"
./calendar.sh delete <calendarId> <eventId> --force

# Search
./calendar.sh search "standup" --max 10 --json

# Free/busy
./calendar.sh freebusy --from 2026-04-24T00:00:00Z --to 2026-04-25T00:00:00Z --json

# RSVP
./calendar.sh respond <calendarId> <eventId> --status accepted

# Workspace helpers
./calendar.sh users --json
./calendar.sh team [e-mail] --json
./calendar.sh conflicts --json

drive.sh

Wraps gog drive. Use for browsing, uploading, downloading, and sharing files.

# List files
./drive.sh ls --json
./drive.sh ls --folder <folderId> --json

# Search
./drive.sh search "Q1 report" --max 10 --json

# Get metadata / URL
./drive.sh get <fileId> --json
./drive.sh url <fileId>

# Download / upload
./drive.sh download <fileId> --out ./local.pdf
./drive.sh upload ./report.pdf --name "Q1 Report" --folder <folderId>

# Manage
./drive.sh copy <fileId> "Copy of Doc"
./drive.sh move <fileId> --folder <targetFolderId>
./drive.sh rename <fileId> "New Name"
./drive.sh delete <fileId> --force
./drive.sh mkdir "Project Assets" --folder <parentFolderId>

# Permissions
./drive.sh share <fileId> --email [e-mail] --role reader
./drive.sh permissions <fileId> --json
./drive.sh unshare <fileId> <permissionId>

docs.sh

Wraps gog docs. Use for reading and writing Google Docs.

# Read
./docs.sh cat <docId>
./docs.sh info <docId> --json
./docs.sh structure <docId>

# Create / copy
./docs.sh create "Meeting Notes"
./docs.sh copy <docId> "Copy of Doc"

# Write
./docs.sh write <docId> --body "Full replacement content"
./docs.sh insert <docId> --body "Inserted text" --index 1
./docs.sh find-replace <docId> "old" "new"
./docs.sh sed <docId> 's/foo/bar/g'
./docs.sh clear <docId> --force

# Export
./docs.sh export <docId> --format txt --out ./doc.txt
./docs.sh export <docId> --format md --out ./doc.md
./docs.sh export <docId> --format pdf --out ./doc.pdf

sheets.sh

Wraps gog sheets. Use for reading and writing Google Sheets.

# Read a range (A1 notation)
./sheets.sh get <spreadsheetId> "Sheet1!A1:D10" --json

# Metadata
./sheets.sh metadata <spreadsheetId> --json

# Write
./sheets.sh update <spreadsheetId> "Sheet1!A1:B2" --values-json '[["Name","Score"],["Alice","95"]]' --input USER_ENTERED
./sheets.sh append <spreadsheetId> "Sheet1!A:C" --values-json '[["2026-04-23","item","10"]]'
./sheets.sh clear <spreadsheetId> "Sheet1!A2:Z" --force

# Find/replace
./sheets.sh find-replace <spreadsheetId> "old" "new"

# Structure
./sheets.sh create "New Spreadsheet"
./sheets.sh add-tab <spreadsheetId> "Tab2"
./sheets.sh rename-tab <spreadsheetId> "Tab2" "Summary"
./sheets.sh delete-tab <spreadsheetId> "OldTab" --force

# Export
./sheets.sh export <spreadsheetId> --format xlsx
./sheets.sh export <spreadsheetId> --format csv

photos.sh

Calls the Google Photos Library REST API using gog OAuth tokens. Unlike other wrappers, this uses curl directly (gog has no native photos command).

One-time setup — re-authorize with photos scopes:

gog auth add [e-mail] \
  --extra-scopes "https://www.googleapis.com/auth/photoslibrary.readonly,https://www.googleapis.com/auth/photoslibrary" \
  --force-consent

# Browse
./photos.sh list --max 10 --json
./photos.sh search "sunset" --max 20 --json
./photos.sh get <mediaItemId> --json
./photos.sh download <mediaItemId> --out ./photo.jpg

# Albums
./photos.sh albums --json
./photos.sh album <albumId> --max 50 --json
./photos.sh create-album "Trip 2026"

# Upload
./photos.sh upload ./photo.jpg
./photos.sh upload ./photo.jpg --album <albumId>

photos-to-drive.sh

Bulk-uploads a local folder of photos/videos to a Google Drive folder. Designed for saving a downloaded Google Photos shared album to Drive.

Workflow:

1. In Google Photos, open the shared album → select all → three-dot menu → Download (saves a zip)
2. Unzip: unzip ~/Downloads/Photos*.zip -d ~/Downloads/album
3. Upload to Drive:

# Create a new Drive folder and upload into it
./photos-to-drive.sh ~/Downloads/album --album-name "Beach Trip 2026"

# Upload into an existing Drive folder
./photos-to-drive.sh ~/Downloads/album --folder <driveId>

# Preview without uploading
./photos-to-drive.sh ~/Downloads/album --album-name "Trip" --dry-run

analyze-for-auction.sh

Analyzes photos in a Google Drive folder using a local Ollama vision model, groups identical items into lots, researches market prices (eBay, Etsy, comparable auctions), and writes everything to a Google Sheet.

Requirements:

• Ollama running with a vision model pulled (e.g. llava)
• Photos already uploaded to Drive (use photos-to-drive.sh first)

One-time model pull (on the Ollama server):

ollama pull llava              # ~4GB, fast
ollama pull llama3.2-vision    # ~8GB, higher quality

# Basic usage (Ollama on localhost)
./analyze-for-auction.sh kellystore4242026

# Custom sheet name
./analyze-for-auction.sh kellystore4242026 --sheet-name "Kelly Store April 2026"

# Ollama on a remote server (standard port 11434)
./analyze-for-auction.sh kellystore4242026 \
  --ollama-host http://openclaw:11434 \
  --model llava \
  --sheet-name "Kelly Store April 2026"

# Dry run — analyze photos and print lots without creating a sheet
./analyze-for-auction.sh kellystore4242026 --dry-run

# Resume after interruption (skips re-analyzing photos)
./analyze-for-auction.sh kellystore4242026 --resume ./analysis_kellystore4242026.json

Output spreadsheet columns: Lot #, Category, Item Name, Brand, Qty, Condition, Description, eBay Low/High, Etsy Low/High, Other Markets, Rec. Low/High, Pricing Notes, Keywords, Photos

Checkpoints: Raw photo analysis is saved to ./analysis_<folder>.json after step 1. If the script is interrupted during grouping/pricing, rerun with --resume to skip straight to the sheet creation.

Patterns

Pipe JSON to jq

./gmail.sh search 'is:unread' --json --results-only | jq -r '.[].id'

Collect IDs then batch-act

ids=$(./gmail.sh search 'label:newsletters older_than:7d' --json --results-only | jq -r '.[].id')
./gmail.sh archive $ids

Dry-run before a destructive action

./drive.sh delete <fileId> --dry-run
./drive.sh delete <fileId> --force

Non-interactive / CI

./calendar.sh create primary --summary "Deploy" --from ... --to ... --no-input --json