How often will this skill prompt me to compact?

By default, the skill suggests a check-in after 50 tool calls, and every 25 calls thereafter. This threshold can be customized using the COMPACT_THRESHOLD environment variable.

Why use manual compaction instead of Claude's automatic system?

Automatic compaction can trigger at arbitrary points, often in the middle of a complex task, causing Claude to lose track of local variables or specific logic. Manual compaction allows you to clear history at logical milestones where context is less critical.

Does this skill work with other AI agents like Cursor or OpenCode?

While specifically optimized for Claude Code's hook system, the strategic principles of milestone-based context clearing apply to any LLM-based agent harness that supports manual context pruning.

What information is lost during a strategic compaction?

Compaction clears the granular conversation history and intermediate reasoning steps. However, it preserves persistent elements like CLAUDE.md, TodoWrite tasks, memory files, and the actual files on your disk.

Strategic Context Compaction

Name: Strategic Context Compaction
Author: affaan-m

byaffaan-m

•

156,033

•

生产力与工作流

Optimizes Claude's context management by suggesting manual compaction at logical task boundaries to prevent critical data loss.

Strategic Context Compaction is a performance optimization skill for Claude Code that replaces unpredictable automatic compaction with intelligent, milestone-based prompts. By monitoring tool usage and session length, it identifies logical transitions—such as moving from research to implementation—where a manual `/compact` will clear bulky reasoning history without losing essential project plans or file states. This ensures Claude remains responsive and accurate during long, complex development sessions exceeding 200K tokens, preventing the 'context pressure' that leads to hallucinations or slow response times.

主要功能

01Seamless integration with PreToolUse hooks for Edit and Write actions

02Context preservation guidance for critical project state

03Transition-aware suggestions (Research → Plan → Implement)

04156,033 GitHub stars

05Customizable notification thresholds for strategic intervention

06Intelligent tool-call tracking to monitor session density

使用场景

01Managing high-token sessions during large-scale codebase refactoring or migrations.

02Clearing session noise after a deep debugging dive before starting a new feature.

03Ensuring project plans are preserved while flushing out verbose file exploration logs.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code strategic-compact

For use in Claude.ai and ChatGPT

主要功能

01Seamless integration with PreToolUse hooks for Edit and Write actions

02Context preservation guidance for critical project state

03Transition-aware suggestions (Research → Plan → Implement)

04156,033 GitHub stars

05Customizable notification thresholds for strategic intervention

06Intelligent tool-call tracking to monitor session density

使用场景

01Managing high-token sessions during large-scale codebase refactoring or migrations.

02Clearing session noise after a deep debugging dive before starting a new feature.

03Ensuring project plans are preserved while flushing out verbose file exploration logs.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code strategic-compact

For use in Claude.ai and ChatGPT