SweetClaude Retro

Parses improvement registers in JSONL or Markdown and generates checkpoint summaries with archived stale entries.

6|3|Updated Apr 12, 2026

Search Tags:#workflow #maintenance #improvement #traceability #project-management #retro

Ready to run

npx skills add https://github.com/carson-sweet/sweetclaude --skill sweetclaude-retro

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: SweetClaude Retro
Source: https://github.com/carson-sweet/sweetclaude/tree/main/skills/retro
Command: npx skills add https://github.com/carson-sweet/sweetclaude --skill sweetclaude-retro

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill solves the problem of project drift and forgotten learnings by providing a structured, repeatable process to review, refine, and archive improvement registers.

Core Features & Use Cases

Improvement Register Management: Automatically parses and weights project learnings to surface the most relevant insights.
Interactive Review: Guides the user through a one-by-one confirmation pass to retire stale patterns or update existing preferences.
Checkpointing: Generates a summary of project health and updates the state to ensure continuous improvement.

Quick Start

Invoke the retro skill to review your current improvement register and update your project state.

Frequently Asked Questions about SweetClaude Retro

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I manage project improvement registers to prevent knowledge drift?▼

Project improvement registers are managed by parsing JSONL or Markdown entries and applying decay-weighted relevance to surface the most pertinent development learnings. This structured process prevents project drift by systematically reviewing and refining registered insights.

What is the best way to run a structured retrospective for software development patterns?▼

A structured retrospective is facilitated through an interactive review session that guides users one-by-one to confirm, retire, or update development patterns and preferences. This ensures continuous project health by evaluating decay-weighted insights.

How does decay-weighted relevance work when reviewing project learnings?▼

Decay-weighted relevance evaluates project learnings by parsing JSONL or Markdown entries to reduce the priority of older, potentially stale patterns over time. This mechanism surfaces the most relevant insights for an interactive review session.

Can I use Markdown and JSONL formats to maintain my project's improvement register?▼

Markdown and JSONL formats are fully supported for maintaining improvement registers, as the system parses these entries to apply decay-weighted relevance. This allows flexible logging of development patterns and preferences.

How do I archive stale development patterns and generate a project health summary?▼

Stale development patterns are archived and a project health summary is generated by creating a checkpoint during the interactive review session. This updates the project state to ensure consistency and continuous improvement.

When should I use a structured retro process for project maintenance?▼

A structured retro process should be used when project drift occurs and forgotten learnings need a repeatable process for review, refinement, and archiving. It is ideal for maintaining long-term project health and state consistency.