Add DB helper scripts for atomic database updates

[babkinvanya]

Summary

• read-db.py — loads all 6 learner databases in one call (replaces 6 Read tool calls at session start). Returns combined JSON with computed fields (due_reviews_count, next_session_id, streak_active).
• update-db.py — atomically updates all 6 databases from a single JSON session report via stdin (replaces 10+ Edit tool calls at session end). Implements SM-2 spaced repetition algorithm, streak tracking, mastery calculation, and error pattern management.
• CLAUDE.md — documents script usage with input schema and examples so Claude knows to use scripts instead of manual edits.
• .gitignore — adds *.json.tmp for atomic write temp files.

Why

After each learning session, Claude was manually editing 6 JSON files with 10+ sequential Edit calls — often taking longer than the lesson itself. These scripts reduce it to a single Bash call each for reading and writing, with atomic writes (.tmp + rename) and automatic pre-update backups.

Usage

# Session start: load all data
python3 .claude/hooks/read-db.py

# Session end: save results
python3 .claude/hooks/update-db.py <<'EOF_DATA'
{ "session_id": "002", "date": "2026-04-02", "duration_minutes": 20, ... }
EOF_DATA

Test plan

• read-db.py outputs valid JSON with all 6 databases and computed fields
• update-db.py correctly updates all 6 files from a test session payload
• Atomic writes confirmed (.tmp → rename)
• Pre-update backup created automatically
• Used in a real learning session (session 002) — all databases updated correctly in one call

🤖 Generated with Claude Code