Why use manual compaction instead of Claude's auto-compaction?

Auto-compaction often triggers at arbitrary token limits, which can happen mid-task and cause Claude to lose sight of your current implementation plan. Manual compaction lets you clear the 'clutter' at logical stopping points.

Does this skill run the /compact command for me?

No, it acts as a strategic advisor. It suggests the best time to compact, but leaves the final decision to you to ensure no active working context is lost prematurely.

Can I customize the frequency of the suggestions?

Yes, you can configure the COMPACT_THRESHOLD environment variable to set the exact number of tool calls that trigger the first suggestion.

How does the skill know when to suggest a compaction?

The skill monitors your session and counts 'Edit' and 'Write' tool calls. Once you reach a threshold (default 50 calls), it provides a helpful reminder to use the /compact command.

Strategic Context Compactor

Name: Strategic Context Compactor
Author: Aventerica89

byAventerica89

0•

Productividad y Flujo de Trabajo

Optimizes AI coding sessions by suggesting manual context compaction at logical task boundaries instead of arbitrary automated triggers.

The Strategic Compact skill enhances long-running development sessions by tracking tool usage and prompting users to manually compact their context at strategic intervals. By monitoring 'Edit' and 'Write' operations, it prevents the loss of critical information that often occurs during Claude's arbitrary auto-compaction. This skill allows developers to preserve implementation plans and research while clearing out stale exploration data, ensuring the AI remains efficient, accurate, and contextually aware throughout complex multi-step coding operations.

Características Principales

01Intelligent tracking of Edit and Write tool invocations

02PreToolUse hook integration for seamless background monitoring

03Logical boundary management to preserve planning context before execution

04Configurable suggestion thresholds to match project complexity

050 GitHub stars

06Periodic reminders to prevent context bloat during extended sessions

Casos de Uso

01Maintaining high response quality during large-scale, multi-file refactoring tasks

02Transitioning from the research and planning phase to the implementation phase

03Clearing trial-and-error debugging context after successfully resolving an error

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add aventerica89/bricks-builder-agent strategic-compact

For use in Claude.ai and ChatGPT

Download Skill