samwang0723/strategic-compact

Strategic Compact

Suggest manual /compact at logical workflow boundaries to preserve important context through task phases, rather than relying on auto-compaction that triggers at arbitrary points mid-task.

Workflow

Step 1: Assess Compaction Need

Check these indicators:

| Indicator | Action | |-----------|--------| | 50+ tool calls in session (hook fires) | Evaluate current phase before compacting | | Transitioning from research → implementation | Compact — research context is bulky, plan is the distilled output | | Just finished debugging a complex issue | Compact — debug traces pollute context for next task | | Mid-implementation with active changes | Do NOT compact — losing file paths, variable names, and partial state is costly | | Switching to an unrelated task | Compact — clear stale context before new focus |

Step 2: Apply the Decision Guide

| Phase Transition | Compact? | Reason | |-----------------|----------|--------| | Research → Planning | Yes | Research context is bulky; plan is the distilled output | | Planning → Implementation | Yes | Plan is in TodoWrite or a file; free up context for code | | Implementation → Testing | Maybe | Keep if tests reference recent code; compact if switching focus | | Debugging → Next feature | Yes | Debug traces pollute context for unrelated work | | Mid-implementation | No | Losing variable names, file paths, and partial state is costly | | After a failed approach | Yes | Clear the dead-end reasoning before trying a new approach |

Step 3: Preserve Before Compacting

Before running /compact, save critical context:

1. Write important findings to files or ~/.claude/memory/
2. Ensure the TodoWrite task list reflects current state
3. Commit any in-progress work to git
4. Add a summary message: /compact Focus on implementing auth middleware next

What Survives Compaction

| Persists | Lost | |----------|------| | CLAUDE.md instructions | Intermediate reasoning and analysis | | TodoWrite task list | File contents previously read | | Memory files (~/.claude/memory/) | Multi-step conversation context | | Git state (commits, branches) | Tool call history and counts | | Files on disk | Nuanced user preferences stated verbally |

Hook Setup

The suggest-compact.sh script (bundled with this skill) tracks tool calls and suggests compaction at configurable thresholds. Add to ~/.claude/settings.json:

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [{ "type": "command", "command": "~/.claude/skills/strategic-compact/suggest-compact.sh" }]
      }
    ]
  }
}

Configure via environment variable: COMPACT_THRESHOLD (default: 50 tool calls before first suggestion, then every 25 calls after).

Context Optimization Tips

• Keep CLAUDE.md lean — it loads into every conversation
• Watch for duplicate rules — same content in ~/.claude/rules/ and project .claude/rules/
• Each loaded skill adds 1-5K tokens — disable unused skills to free context
• Large tool results (file reads, searches) accumulate fast — compact after exploration-heavy phases