continuous-learning

Extract reusable patterns from Gemini CLI session transcripts and save them as learned skills.

86|21|Updated Feb 9, 2026
One-click install
npx skills add https://github.com/Jamkris/everything-gemini-code --skill continuous-learning-jamkris
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: continuous-learning
Source: https://github.com/Jamkris/everything-gemini-code/tree/main/skills/continuous-learning
Command: npx skills add https://github.com/Jamkris/everything-gemini-code --skill continuous-learning-jamkris

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq.

What problem does it solve?

Gemini CLI sessions yield valuable interaction patterns, but manual review is time-consuming; this skill automates extraction and persistence of reusable patterns as learned skills.

Core Features & Use Cases

  • Stop-hook style evaluation at session end to detect patterns such as error resolutions, user corrections, workarounds, and debugging techniques.
  • Persists extracted patterns to ~/.gemini/skills/learned/ for reuse in future sessions.
  • Supports configuration via config.json to adjust thresholds and paths for automation.

Quick Start

Run the continuous-learning evaluate-session.sh script at the end of a Gemini session to publish learned patterns to the designated directory.

Frequently Asked Questions about continuous-learning

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

FAQPage Schema
How do I extract reusable patterns from Gemini CLI session transcripts?

You can extract reusable patterns from Gemini CLI sessions by running an evaluate-session.sh script at session end, which detects error resolutions, workarounds, and debugging techniques to save as learned skills.

Can I automatically save debugging techniques and error resolutions from Gemini sessions?

Yes, automating the extraction of debugging techniques and error resolutions is possible by applying a Stop hook at the end of Gemini sessions to evaluate transcripts and persist detected patterns.

What is the minimum session length needed to trigger continuous learning pattern detection?

Continuous learning pattern detection requires a Gemini session length of 10 or more messages to evaluate the transcript and extract reusable patterns like project-specific conventions and user corrections.

Do I need jq installed to automate saving learned skills from Gemini sessions?

Yes, you need jq installed as a dependency to run the evaluate-session.sh script, which generates learned skills and stores them in the ~/.gemini/skills/learned directory.

How do I configure the threshold for extracting workarounds from Gemini session transcripts?

You can configure thresholds and paths for extracting workarounds by editing a config.json file, which controls how the evaluate-session.sh script processes and persists session patterns.

Where are detected user corrections and project conventions stored for future Gemini sessions?

Detected user corrections and project conventions are stored as learned skills in the ~/.gemini/skills/learned directory, enabling their reuse in future Gemini CLI sessions.