update

README.md

@@ -21,24 +21,36 @@ This scrapes unchecked URLs from `scraping.md`, saves JSON to `scraped_data/`, a
 
 ### 2. Process scraped data into docs/
 
-```bash
-# Process 1 file (default)
-venv/bin/python python/process.py
+Use this prompt with Claude Code to process pending JSON files into documentation:
 
-# Process 5 files
-venv/bin/python python/process.py -n 5
+```
+Process 20 pending JSON files from processed.md into docs/.
 
-# Process all remaining files
-venv/bin/python python/process.py -n 9999
+OUTER LOOP (repeat until X files total are done):
+  1. Read processed.md, find next 5 pending files (marked `- [ ]`)
+  2. Mark each in-progress: `- [ ]` to `- [~]`
 
-# Process in parallel (e.g., 4 workers processing 10 files each)
-for i in {1..4}; do venv/bin/python python/process.py -n 10 & done; wait
+  INNER LOOP (process batch of 5 in parallel):
+    3. Spawn 5 Task sub-agents in ONE message (subagent_type="general-purpose", run_in_background=true), one per file:
+
+       "Process scraped_data/{filename} into docs/:
+       - Read the JSON
+       - Translate: Liberti, Plebejer=Plebeians, Equites, Patrizier=Patricians, Wanderer=Waders, Schmiede=Smiths, Älteste=Elders, Mercatoren=Mercators, Edelmänner=Nobles
+       - Target: anno-117-buildings_* → docs/buildings/, anno-117-goods_* → docs/goods/, anno-117-specialists* → docs/specialists/, anno-117-skills_* → docs/skills/
+       - Merge if exists, create if not (use existing docs/ as format examples)
+       - Update category _index.md if needed
+       - Mark done: `- [~] {filename}` to `- [x] {filename}` in processed.md"
+
+    4. Use TaskOutput to wait for all 5 agents to complete
+
+  END INNER LOOP
+
+  5. Count completed files, continue OUTER LOOP until X total done
+
+END OUTER LOOP
 ```
 
-The script uses file locking to safely run in parallel. Each invocation:
-1. Claims pending JSON files from `processed.md`
-2. Calls Claude to parse them into the `docs/` folder structure
-3. Marks them as completed
+Replace X with desired total (e.g., 20).
 
 ---
 
@@ -49,7 +61,6 @@ The script uses file locking to safely run in parallel. Each invocation:
 - `scraped_data/` - Raw scraped JSON files
 - `docs/` - Structured documentation (see CLAUDE.md for structure)
 - `python/scraper.py` - Web scraper script
-- `python/process.py` - Process one JSON file into docs/
 
 ## Data Flow
 
@@ -57,6 +68,6 @@ The script uses file locking to safely run in parallel. Each invocation:
 anno.land pages
       ↓ (scraper.py)
 scraped_data/*.json
-      ↓ (process_one.py → Claude)
+      ↓ (Claude sub-agents)
 docs/**/*.md
 ```