kramme:learn

Add learnings to a SQLite database and propose AGENTS.md updates.

2|2|Updated Jan 15, 2026
One-click install
npx skills add https://github.com/Abildtoft/kramme-cc-workflow --skill kramme-learn
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kramme:learn
Source: https://github.com/Abildtoft/kramme-cc-workflow/tree/main/skills/kramme%3Alearn
Command: npx skills add https://github.com/Abildtoft/kramme-cc-workflow --skill kramme-learn

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires sqlite3, and includes scripts (resource) components.

What problem does it solve?

Add a learning to the persistent SQLite database at ~/.kramme-cc-workflow/learnings.db. Proposes AGENTS.md updates if the learning is project-relevant.

Core Features & Use Cases

  • Persist learnings to the local learnings database to centralize project knowledge.
  • Support both Quick Mode interactive entry and guided input for adding learnings.
  • Propose adding relevant learnings to AGENTS.md to improve project documentation and onboarding.

Quick Start

/kramme:learn "Always check for null" --category Quality

Frequently Asked Questions about kramme:learn

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

FAQPage Schema
How do I persist project learnings to a SQLite database from the command line?

To log learnings to a database, this skill inserts records into a persistent SQLite database using Bash scripts and sqlite3. It sanitizes inputs and stores project knowledge centrally at a local database file path.

What is the best way to automatically propose AGENTS.md updates from new learnings?

Proposing AGENTS.md updates involves evaluating new learnings for project relevance and generating a formatted text proposal. This skill checks if the project warrants an update and formats the learning for direct addition to the file.

Do I need sqlite3 installed to log learnings to a local database?

Yes, sqlite3 is required. The skill depends on sqlite3 and Bash scripts to handle database inserts, sanitize user inputs, and manage the persistent local learnings database file.

Can I use quick mode for interactive entry of bash workflow learnings?

Yes, Quick Mode supports interactive entry of bash workflow learnings. You can pass a direct text string and category flag for fast input, or use the guided input mode to add entries to the SQLite database step by step.

How does this approach compare to manually editing AGENTS.md for project knowledge?

Compared to manually editing AGENTS.md, this approach centralizes project knowledge in a SQLite database first. It then selectively proposes formatted updates to the markdown file, ensuring learnings are sanitized and persistently stored before documentation.