Where is the captured knowledge stored?

Learnings are stored as individual skills within the .claude/skills/ directory of your project root, indexed by a central registry for easy retrieval.

How do I trigger the learning process?

You can use the /learn command or ask Claude to 'remember' a specific pattern, solution, or 'gotcha' from your current conversation.

What is Investigation Mode?

If context is lacking, Sparks Learn deploys subagents to explore your codebase, map files, and verify how a feature or pattern works before documenting it.

What is Sparks Learn?

It is a Claude Code skill that captures and organizes project-specific knowledge like patterns, decisions, and procedures into a reusable format for AI assistants.

Does it update existing skills?

Yes, it proactively offers to update previously loaded skills if it discovers new information, fixes, or better patterns during the current session.

Sparks Learn

Name: Sparks Learn
Author: Codename-Inc

byCodename-Inc

0•

Learning & Documentation

Captures and preserves project-specific knowledge, architectural decisions, and coding patterns to build a durable codebase memory for AI assistants.

Sparks Learn is a specialized knowledge-capture tool for the Claude Code environment that transforms fleeting conversation insights into permanent, actionable skills. It proactively identifies learning opportunities—such as hard-won debugging insights, architectural rationale, and complex procedures—and stores them in a structured registry. By automating the creation of project-specific documentation, it ensures that any team member, human or AI, can become productive immediately without losing context or repeating past mistakes. Through its unique Investigation Mode, it can even explore unfamiliar codebases to generate high-fidelity documentation from scratch.

Key Features

010 GitHub stars

02Automated registry management for easy discovery of existing project learnings

03Proactive skill updates to refine existing knowledge during active sessions

04Standardized templates for features, gotchas, patterns, and architectural decisions

05Context-aware parsing of conversation history to identify high-value insights

06Investigation Mode using subagents to map and analyze unfamiliar codebases

Use Cases

01Recording the rationale behind a complex architectural decision for future maintainers

02Documenting a 'gotcha' after a long debugging session to prevent future occurrences

03Creating 'how-to' guides for multi-step deployment or release procedures within the CLI

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add codename-inc/spectre-labs sparks-learn

For use in Claude.ai and ChatGPT

Download Skill