Where are the research findings stored?

Findings are saved as dated Markdown files in a designated inbox folder (default is /inbox) defined in your derivation-manifest.md.

What search engines does the Learn skill support?

The skill uses a tool cascade starting with Exa Deep Researcher and Exa Web Search, falling back to standard WebSearch tools if the primary MCPs are unavailable.

How do I specify the depth of the research?

You can force a specific depth by adding --light, --moderate, or --deep to your command, such as '/learn AI security --deep'.

Can this skill help manage my project objectives?

Yes, it automatically reads your goals.md file to find research directions and appends new topics discovered during the research process.

What is 'provenance frontmatter'?

It is YAML metadata at the top of the research file that records exactly how the information was gathered, including the tool used, the exact prompt, and the research ID for auditing.

Deep Knowledge Researcher

Name: Deep Knowledge Researcher
Author: micsapp

bymicsapp

0•

学習とドキュメンテーション

Conducts multi-stage research and builds a structured knowledge graph with full source provenance.

The Learn skill is a sophisticated research agent designed for Claude Code that automates the discovery and synthesis of complex information. By leveraging advanced tools like Exa Deep Researcher and web search cascades, it investigates topics at varying depths—from quick fact-checks to comprehensive deep dives. Every research session produces structured Markdown files in a local inbox, complete with detailed YAML frontmatter for provenance tracking. This skill is the primary entry point for building a persistent knowledge graph, automatically identifying follow-up questions and updating project goals to ensure continuous learning and architectural clarity.

主な機能

010 GitHub stars

02Multi-tier research cascade using Exa Deep Research, Exa Web Search, and standard WebSearch.

03Seamless pipeline chaining to downstream skills like /reduce for insight extraction and synthesis.

04Intelligent goal management that reads and updates goals.md with discovered research directions.

05Automated provenance tracking with metadata including original prompts, tool IDs, and timestamps.

06Configurable research depth (light, moderate, deep) via command-line flags or project configuration.

ユースケース

01Automating the collection of documentation and practical implementation patterns for specific libraries.

02Conducting deep-dive competitive analysis or feature research for product development.

03Exploring new technical domains or architectural patterns and filing them into a project knowledge base.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add micsapp/micstec-skills learn

For use in Claude.ai and ChatGPT