How much token space does the memory-lite feature save?

The tiered loading strategy (Core, Summary, and Detail layers) reduces startup token consumption from over 40% to less than 10%, providing significantly more room for actual coding and analysis.

How does the seamless switching work?

When token usage hits the 60% threshold, the skill automatically saves the current state to memory files and triggers an agentTurn to create a new session, loading only essential data to keep the conversation flowing.

Will I lose progress when a session switches?

No. The skill automatically captures task status, key decisions, and progress in MEMORY.md and daily logs, ensuring the new session picks up exactly where the last one left off without user intervention.

Can I customize the token threshold?

Yes, users can modify the DIALOG_THRESHOLD in the configuration scripts to trigger transitions at specific usage levels suited for their particular LLM model limits.

Context Manager

Name: Context Manager
Author: dvcrn

bydvcrn

•

生产力与工作流

Automates session switching and memory persistence to manage long-running Claude conversations without losing project context.

Context Manager (v2) is a specialized skill designed to overcome context window limitations by automatically monitoring token usage and performing seamless session transitions. It utilizes a sophisticated tiered memory architecture—separating core identity, lite summaries, and detailed logs—to reduce startup token overhead by up to 75%. By integrating real-time API monitoring and automated memory persistence, this tool ensures the AI retains critical project states and task progress across session boundaries without requiring any manual user intervention.

主要功能

01Automated memory synchronization across MEMORY.md and daily activity logs

02Tiered memory loading architecture to reduce startup token usage by 75%

03Zero-intervention 'agentTurn' mechanism for continuous AI workflows

04Real-time API-based monitoring of total and context tokens

053 GitHub stars

06Automatic seamless session switching triggered at 60% token capacity

使用场景

01Optimizing token efficiency in long-form research and documentation sessions

02Maintaining continuity during multi-day coding projects or complex software architecture tasks

03Automating the capture of project decisions and task states across multiple AI sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dvcrn/openclaw-skills-marketplace context-manager

For use in Claude.ai and ChatGPT

主要功能

01Automated memory synchronization across MEMORY.md and daily activity logs

02Tiered memory loading architecture to reduce startup token usage by 75%

03Zero-intervention 'agentTurn' mechanism for continuous AI workflows

04Real-time API-based monitoring of total and context tokens

053 GitHub stars

06Automatic seamless session switching triggered at 60% token capacity

使用场景

01Optimizing token efficiency in long-form research and documentation sessions

02Maintaining continuity during multi-day coding projects or complex software architecture tasks

03Automating the capture of project decisions and task states across multiple AI sessions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dvcrn/openclaw-skills-marketplace context-manager

For use in Claude.ai and ChatGPT