rrr

Generate session retrospectives, AI diaries, and lessons learned files.

1|Updated May 24, 2026
One-click install
npx skills add https://github.com/tinner-deinno/innova-skills-lib --skill rrr-tinner-deinno
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rrr
Source: https://github.com/tinner-deinno/innova-skills-lib/tree/main/core/rrr
Command: npx skills add https://github.com/tinner-deinno/innova-skills-lib --skill rrr-tinner-deinno

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

rrr turns a plain end-of-session “wrap up” into a structured retrospective that captures decisions, timeline, friction, and lessons learned so you improve over time instead of forgetting what happened.

Core Features & Use Cases

  • Session retrospective with AI diary: writes an end-to-end retro including a first-person diary, honest feedback, and next steps.
  • Timeline from real user-message timestamps: in default mode it runs a background timestamp miner against the session .jsonl to avoid inventing times.
  • Lesson learning + pattern review: saves a generalizable lesson file and appends a required metrics row, optionally flagging recurring friction/error themes.

Quick Start

Ask your AI agent: "rrr --detail" to generate a full session retrospective with an accurate timeline and detailed diary.

Frequently Asked Questions about rrr

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

FAQPage Schema
How do I generate a session retrospective after finishing a coding task?

To generate a session retrospective, invoke the rrr command to create a structured wrap-up capturing decisions, timeline, friction, and lessons learned. It mines real timestamps from your local session logs to produce an accurate end-to-end retrospective.

Can I create an AI diary from my Claude Code session history?

Yes, you can create an AI diary by running rrr in default or --detail mode. It produces a first-person diary entry alongside honest feedback and next steps by mining timestamps directly from the local Claude Code .jsonl session file.

What is the best way to capture lessons learned from a development workflow?

The best way to capture lessons learned is using the rrr --detail command, which saves a generalizable lesson file and appends a required metrics row to your memory vault. It optionally flags recurring friction and error themes for pattern review.

Does the retrospective tool work without full session logs?

Yes, the retrospective tool works without full logs by using the --quick memory-only mode for a faster wrap-up. When session .jsonl files are available, it runs a background timestamp miner to avoid inventing times.

When should I use multi-subagent mode for a session summary?

You should use the --deep multi-subagent mode for a session summary when you need a highly comprehensive retrospective with extensive analysis. This mode provides the deepest level of processing compared to the default, --quick, and --detail options.