Agent Skills: synchronizing-grok-convos

Global/continuous sync of your private Grok conversations from both grok.com and x.com into a clean Obsidian vault. Uses Playwright for reliable discovery across surfaces, supports bookmarklet capture, requires Defuddle, and produces 6-stream maximized Markdown with surface origin in frontmatter. Default target: ~/.grok/conversations/

UncategorizedID: m31uk3/ai-skills/synchronizing-grok-convos

Install this agent skill to your local

pnpm dlx add-skill https://github.com/m31uk3/ai-skills/tree/HEAD/skills/synchronizing-grok-convos

Skill Files

Browse the full folder contents for synchronizing-grok-convos.

Download Skill

Loading file tree…

skills/synchronizing-grok-convos/SKILL.md

Skill Metadata

Name
synchronizing-grok-convos
Description
"Global/continuous sync of your private Grok conversations from both grok.com and x.com into a clean Obsidian vault. Uses Playwright for reliable discovery across surfaces, supports bookmarklet capture, requires Defuddle, and produces 6-stream maximized Markdown with surface origin in frontmatter. Default target: ~/.grok/conversations/"

synchronizing-grok-convos

Robust global sync for Grok conversations across grok.com and x.com.

All conversations land in a single folder (~/.grok/conversations/ by default). The surface is recorded in frontmatter so you can filter easily.

CLI (recommended after build)

# One-time build of the ggs binary
bun install
bun build ./bin/ggs.ts --compile --outfile bin/ggs

# Global sync (uses default ~/.grok/conversations/)
./bin/ggs --sync --headed     # first run for login

# One specific conversation
./bin/ggs "https://x.com/i/grok?conversation=2068028429640024253"

# Custom target
./bin/ggs --sync --target ~/path/to/vault/GenAI/Grok

Inside Grok TUI

/synchronizing-grok-convos --sync
/synchronizing-grok-convos <url-or-id>

Core Features

  • Dual-surface discovery (grok.com + x.com/i/grok)
  • Incremental sync via .grok-sync-index.json in target
  • Hybrid capture: Playwright or --html (bookmarklet)
  • Defuddle required for cleaning + metadata
  • 6-stream output with proper priority sections
  • surface frontmatter key for origin tracking
  • Obsidian-native (status, word_count, kebab-friendly ids, etc.)

Bookmarklet (first-class capture)

See assets/grok-export-bookmarklet.js (or scripts/).

  1. Create a bookmark with the JS.
  2. On any open Grok conversation page, click it.
  3. Feed the resulting JSON with --html.

References

  • references/sources.md — full research, X thread, Dev process slugs
  • references/harnesses/ — cron, launchd, TUI schedule examples
  • Architecture diagrams and 6-stream spec in references/

First-time login

ggs --sync --headed

Subsequent runs can be headless.


This skill was built from the egc--grok-export-defuddle research (see references for the full trail including the June 2026 feature request thread).