x-grok-to-obsidian

Export Grok conversations in two stages.

## Stage 1 — Capture conversation JSON from X/Grok

Run the browser script in Chrome DevTools Console on `https://x.com/i/grok` while logged in.

Use script: `scripts/export_grok_items_capture.js`

Behavior:
- Intercept only `GrokConversationItemsByRestId` responses (`fetch` + `XHR`)
- Load chat history with multi-pass scrolling
- Open each discovered conversation to trigger backend responses
- Save one JSON file with conversation metadata + ordered item payload

Quick settings (edit at top of script before run):
- `INDEX_PASSES` (default `3`)
- `CAPTURE_PASSES` (default `3`)
- `MAX_CHATS` (`null` = all, or number for test)

Output:
- `grok-network-capture-<timestamp>.json` downloaded by browser

## Stage 2 — Convert JSON to Obsidian Markdown

Run Python converter locally:

```bash
python3 scripts/convert_grok_capture_to_md.py \
  --input /path/to/grok-network-capture-*.json \
  --out /path/to/output-folder
```

Converter defaults:
- Frontmatter fields: `URL`, `created`
- Body starts immediately with `# <title>` (no blank line before header)
- Turn headings: `## User` / `## Grok` (no numbering)
- Turn separator: `---`
- Turn order: reverse API item order (API is newest-first)
- Reasoning/deepsearch omitted by default

Useful flags:
- `--include-reasoning` include `thinking_trace` blocks
- `--separator "---"` customize turn separator
- `--overwrite` overwrite same-title files instead of creating `Title 2.md`

## Notes

- Prefer several index/capture passes because X history rendering is lazy and inconsistent.
- If discovered chat count is unexpectedly low, re-run Stage 1 with higher pass counts.
- Keep scripts generic; avoid user-specific absolute paths.

# x-grok-to-obsidian

Reusable OpenClaw skill to export Grok conversations from X (x.com) and convert them into Obsidian-ready Markdown files.

## What it does

Two-stage workflow:

1. **Capture JSON from X/Grok** in browser (network-level capture of `GrokConversationItemsByRestId`).
2. **Convert captured JSON to Markdown** with role-separated turns (`User` / `Grok`) and Obsidian frontmatter.

## Skill structure

- `SKILL.md` — skill metadata + usage instructions
- `scripts/export_grok_items_capture.js` — browser capture script (run in DevTools)
- `scripts/convert_grok_capture_to_md.py` — JSON → Markdown converter

## Quick usage

### 1) Capture from X/Grok

- Open `https://x.com/i/grok` while logged in.
- Open browser DevTools Console.
- Paste/run `scripts/export_grok_items_capture.js`.
- Download output JSON: `grok-network-capture-<timestamp>.json`.

### 2) Convert to Markdown

```bash
python3 scripts/convert_grok_capture_to_md.py \
  --input /path/to/grok-network-capture-*.json \
  --out /path/to/output-folder
```

## Converter defaults

- Frontmatter: `URL`, `created`, optional `source_tweets`
- Turn headings: `## User` / `## Grok`
- Turn separator: `---`
- API item order reversed to conversation order
- Reasoning/deepsearch excluded by default

## Notes

- X history loading is lazy; multi-pass discovery improves completeness.
- This project intentionally avoids shipping personal data and auth artifacts.

## Troubleshooting / operational notes

### 1) History count is too low (e.g., ~166 only)

- Increase discovery passes in `export_grok_items_capture.js` (`INDEX_PASSES`).
- Keep the tab focused while indexing.
- Re-run once; X history rendering is inconsistent and can return different visible subsets between runs.

### 2) Script hangs during capture loop

- The script uses per-conversation timeout watchdogs to avoid infinite waits.
- If a specific chat times out, it is retried in later passes.
- Blocking dialogs are auto-closed during waits.

### 3) 429 Too Man...