context-status

# /context-status — Check Session Health Show the current session status including context usage estimate, active plan, and preservation state. ## What This Skill Shows 1. **Context usage estimate** — Approximate % of context window used 2. **Active plan** — Current plan file and status 3. **Session log** — Most recent session log 4. **Preservation state** — What will survive compaction ## Workflow ### Step 1: Check the Context Estimate The context-monitor hook writes its estimate to `context-pct.txt` (and its counters and threshold flags to `context-monitor-cache.json`) in this project's session directory: ```bash d=~/.claude/sessions/$(printf '%s' "${CLAUDE_PROJECT_DIR:-$PWD}" | python3 -c 'import sys,hashlib;print(hashlib.md5(sys.stdin.read().encode()).hexdigest()[:8])') cat "$d/context-pct.txt" 2>/dev/null; cat "$d/context-monitor-cache.json" 2>/dev/null | head -20 ``` ### Step 2: Find Active Plan ```bash ls -lt quality_reports/plans/*.md 2>/dev/null | head -3 ``` ### Step 3: Find Session Log ```bash ls -lt quality_reports/session_logs/*.md 2>/dev/null | head -1 ``` ### Step 4: Report Status Format the output: ``` 📊 Session Status ───────────────────────────────── Context Usage: ~XX% (estimated) Auto-compact: [approaching | not imminent] 📋 Active Plan File: quality_reports/plans/YYYY-MM-DD_description.md Status: [draft | approved | in_progress | completed] Task: [current unchecked task or "none"] 📝 Session Log File: quality_reports/session_logs/YYYY-MM-DD_description.md ✓ Preservation Check • Pre-compact hook: [configured | missing] • Post-compact restore: [configured | missing] • Session state will be saved before compaction ``` ## Notes - Context % comes from Claude Code's own `context_window.used_percentage` where the status line shows it; the hook's `context-pct.txt` holds an estimate from transcript size (falling back to a tool-call count). The built-in `/context` shows exact usage - Actual compaction is triggered by Claude Code automatically - All important state is saved to disk (plans, logs, MEMORY.md)