When should I trigger a context compaction?

Compaction is recommended when your context usage reaches between 75% and 90% of the token limit to prevent session interruption.

Does this skill help reduce API costs?

Yes, by removing redundant tool outputs and compressing historical messages, it reduces the total token count per request, resulting in lower costs and faster response times.

How does Context Compactor decide what to keep?

It uses a scoring strategy that marks architectural decisions, unresolved bugs, and recent file changes as 'must-preserve,' while summarizing completed work and discarding stale search results.

Will I lose critical information during compaction?

No, the tool is specifically designed to protect critical metadata, error patterns, and current feature states to ensure Claude never loses the project's core objectives.

Context Compactor

Name: Context Compactor
Author: adaptationio

byadaptationio

0•

생산성 및 워크플로

Manages and optimizes Claude Code session context by automatically summarizing history and discarding redundant data to stay within token limits.

Context Compactor is an essential context-engineering utility designed to handle long-running AI development sessions where token limits become a constraint. It intelligently analyzes session history using a priority-based scoring system to preserve critical information like architectural decisions, unresolved bugs, and the current feature state while discarding redundant tool outputs and stale search results. This ensures that Claude maintains high-quality reasoning and project continuity without performance degradation or unnecessary token expenditure, making it ideal for complex, multi-stage coding tasks.

주요 기능

01Automated removal of redundant tool outputs and stale search results

02Context validation to ensure session coherency after compaction

03Priority-based preservation of architectural decisions and unresolved blockers

040 GitHub stars

05Multi-tier token budget management with automated warning and critical thresholds

06Intelligent summarization of completed features and resolved error logs

사용 사례

01Optimizing token usage in large-scale code refactoring projects

02Preparing clean, condensed context for session handoffs or memory storage

03Maintaining session continuity during complex, multi-day development tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adaptationio/skrillz context-compactor

For use in Claude.ai and ChatGPT

Download Skill