Why use manual compaction over auto-compaction?

Auto-compaction often triggers at arbitrary points, potentially losing vital context in the middle of a complex task; manual suggestions allow you to compact at logical breaks.

How does this skill track when to suggest a compact?

It monitors the number of 'Edit' and 'Write' tool calls using a PreToolUse hook and alerts you once a configurable threshold (default 50) is met.

Does this skill perform the compaction automatically?

No, it provides a suggestion based on usage patterns, allowing the developer to decide if the current moment is a logical boundary for clearing context.

Can I change the frequency of suggestions?

Yes, you can set the COMPACT_THRESHOLD environment variable to adjust how many tool calls occur before the first suggestion is triggered.

Strategic Context Compaction

Name: Strategic Context Compaction
Author: ysyecust

byysyecust

0•

Productivity & Workflow

Optimizes Claude's context window by suggesting manual compaction at logical task boundaries to prevent mid-task memory loss.

The Strategic Compact Skill enhances Claude Code sessions by replacing arbitrary auto-compaction with intelligent, threshold-based reminders. By monitoring tool usage and providing alerts at configurable intervals, it empowers developers to clear out stale research or debugging context precisely after reaching milestones or before shifting tasks. This ensures that critical information, such as implementation plans and core logic, is preserved while keeping token usage efficient and preventing the AI from losing focus during complex, multi-step operations.

Key Features

01Non-disruptive suggestions that respect logical task boundaries

02Customizable compaction triggers via environment variables

030 GitHub stars

04Automated PreToolUse hooks for Edit and Write operations

05Intelligent threshold detection based on tool call counts

06Periodic reminders to maintain session hygiene during long tasks

Use Cases

01Wiping error logs and debugging history once a bug is successfully resolved

02Maintaining session performance during large-scale refactors by managing token bloat

03Clearing exploration context after finalizing a technical implementation plan

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ysyecust/everything-claude-code strategic-compact

For use in Claude.ai and ChatGPT

Download Skill