Where is the checkpoint data saved?

The skill updates .sessions/SESSION_[name].md for state and progress logs, and adds new insights to the Lessons Inbox in MEMORY.md.

How does this differ from a full session close?

Unlike a full session close, this skill is designed for rapid, multiple uses per cycle, requiring no handoff summary while focusing on raw progress logs.

What is the 150% Checkpoint Rule?

It means capturing 100% of the core state (what changed) plus an additional 50% of enhancements (key learnings and immediate next steps).

Does it log failures as well as successes?

Yes, a core principle of this skill is logging 'What Didn't Work' to provide raw material for extracting long-term lessons and avoiding future dead ends.

When should I trigger a checkpoint?

Recommended triggers include nearing token limits, finishing a sub-feature, or after a significant bug fix, usually every 30-45 minutes.

Mid-Session Progress Checkpoint

Name: Mid-Session Progress Checkpoint
Author: MykhailoDmytriakha

byMykhailoDmytriakha

0•

Productividad y Flujo de Trabajo

Captures rapid state snapshots and developmental insights during active coding sessions to prevent context loss.

This skill implements a robust checkpointing protocol for Claude Code sessions, allowing developers to preserve their current state and lessons learned without performing a full session handoff. It focuses on the '150% Rule'—capturing not just what changed, but the critical 'why' and 'how' behind successful and failed attempts. By logging the journey of development, it ensures that valuable insights aren't lost when token limits are reached or when transitioning between complex tasks, making it an essential tool for long, iterative development cycles.

Características Principales

01Ultra-quick checkpoint format for high-speed workflows

02Comprehensive 'What Worked/What Didn't' progress logging

03Context snapshotting for low-token scenarios

04Automated lesson extraction into MEMORY.md

05Standardized state persistence in local session logs

060 GitHub stars

Casos de Uso

01Preserving progress when the AI context window is nearly full

02Documenting failed experiments and dead ends to avoid repeating mistakes

03Capturing immediate next steps before pausing or switching complex tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mykhailodmytriakha/my-preacher-helper 74-mid-session-save-150

For use in Claude.ai and ChatGPT

Download Skill