What is the 3/3 counterfactual test used by this skill?

It is a quality bar where a fix is only recommended if it points to exact rework, would have triggered before that rework, and would have produced the correct output initially.

Does this skill automatically modify my skill definitions?

No, it generates high-confidence suggestions and recommendations based on session evidence, allowing you to review and apply the improvements manually.

Can I use this for sessions that happened on different projects?

Yes, as long as you provide the correct session ID or the direct path to the .jsonl session file, the skill can analyze the history.

What kind of patterns does the session analyzer look for?

It specifically looks for code iterations (fixing errors), user corrections ('no, I meant...'), skipped workflow steps, and missing upfront questions.

How do I find my Claude Code session ID for analysis?

Session IDs are typically displayed at the start of a session or can be found by listing files in the ~/.claude/projects directory.

Session Analysis & Skill Optimizer

Name: Session Analysis & Skill Optimizer
Author: doodledood

bydoodledood

•

開発者ツール

Analyzes Claude Code sessions to identify workflow failures and generate high-confidence, evidence-based improvements for agent skills.

The learn-from-session skill provides an automated retrospective layer for Claude Code workflows by parsing interaction logs to identify points of friction. By examining session JSONL files, it detects patterns such as repeated iterations, manual user corrections, and skipped planning phases. The skill applies a rigorous '3/3 counterfactual test' to ensure all recommendations are high-signal, meaning the suggested changes are guaranteed to have prevented specific rework if they had been active during the original session. This creates a continuous improvement loop for developers looking to refine their custom agentic workflows.

主な機能

018 GitHub stars

02Automated session log parsing via UUID or file path

03Evidence-based counterfactual analysis for high-signal fixes

04Standardized analysis logging for persistent external memory

05Pattern detection for iterations, user corrections, and workflow deviations

06Automated discovery of relevant SKILL.md and agent definitions

ユースケース

01Optimizing custom agent skills after complex development tasks

02Debugging why specific workflows required manual intervention

03Refining prompt engineering and skill definitions based on real-world performance

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add doodledood/claude-code-plugins learn-from-session

For use in Claude.ai and ChatGPT

Download Skill