How do I customize the frequency of compaction reminders?

You can set the COMPACT_THRESHOLD environment variable to change the number of tool calls that occur before the skill suggests a manual compaction.

Why use manual compaction over Claude's auto-compaction?

Auto-compaction often triggers mid-task, causing Claude to lose critical reasoning. Strategic compaction happens at logical boundaries, ensuring important context is saved before the history is cleared.

When is the best time to compact my session context?

The best times are typically after finalizing a plan, after completing a major milestone, or after a failed approach to clear out dead-end reasoning.

What information is lost during a compaction?

Compaction clears intermediate reasoning, analysis, and tool call history. However, files on disk, memory files, and specific instructions in CLAUDE.md persist.

Strategic Context Compaction

Name: Strategic Context Compaction
Author: saar120

bysaar120

0•

生产力与工作流

Optimizes Claude's memory management by suggesting manual context compaction at logical workflow boundaries to prevent information loss.

Strategic Context Compaction is a workflow utility designed to improve the performance and coherence of long-running Claude Code sessions. By monitoring tool call counts, the skill prompts users to use the /compact command at strategic intervals—such as moving from research to planning or implementation to testing—rather than allowing Claude to trigger arbitrary auto-compaction. This ensures that critical reasoning and implementation details are preserved while bulky, unnecessary history is cleared, maintaining high response quality even in complex, multi-phase development tasks.

主要功能

01Configurable tool-call thresholds via environment variables

020 GitHub stars

03Threshold-based reminders triggered by tool invocation counts

04Detailed decision matrix for evaluating context retention

05Phase-aware compaction guidance for research, planning, and execution

06Seamless integration with PreToolUse hooks for Edit and Write actions

使用场景

01Clearing bulky debugging logs and error traces before pivoting to a new feature

02Ensuring implementation plans are retained while flushing exploratory research data

03Managing context pressure in long development sessions exceeding 200K tokens

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry strategic-compact

For use in Claude.ai and ChatGPT

主要功能

01Configurable tool-call thresholds via environment variables

020 GitHub stars

03Threshold-based reminders triggered by tool invocation counts

04Detailed decision matrix for evaluating context retention

05Phase-aware compaction guidance for research, planning, and execution

06Seamless integration with PreToolUse hooks for Edit and Write actions

使用场景

01Clearing bulky debugging logs and error traces before pivoting to a new feature

02Ensuring implementation plans are retained while flushing exploratory research data

03Managing context pressure in long development sessions exceeding 200K tokens

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry strategic-compact

For use in Claude.ai and ChatGPT