How does the skill detect my project name?

The skill uses a priority system to detect your project, checking the QMD_PROJECT environment variable, the Git remote URL, the Git repository folder name, or the current directory name.

What is the qmd MCP server?

The qmd MCP (Model Context Protocol) server is a tool that allows AI assistants like Claude to interact with Markdown-based knowledge bases using keyword, semantic, and hybrid search capabilities.

Is this skill compatible with tools other than Claude Code?

Yes, it is fully compatible with OpenCode, Amp, Codex, and other AI coding tools that support MCP servers and custom skills.

Where is the project knowledge data stored?

All knowledge files are stored locally on your machine in the ~/.ai-knowledges/ directory, organized into subfolders by project name.

Do I need to manually update the search index?

The included record.sh script automatically runs 'qmd embed' after every entry to ensure new content is immediately searchable. Manual updates are only needed if you edit files directly.

Project Knowledge Manager

Name: Project Knowledge Manager
Author: jellydn

byjellydn

•

学習とドキュメンテーション

Captures and retrieves project-specific learnings, architectural decisions, and issue resolutions using a searchable Markdown-based knowledge base.

Project Knowledge Manager is a specialized skill designed to build institutional memory within software projects by integrating the qmd MCP server. It enables developers and AI assistants to record technical insights, document complex issue resolutions, and maintain project conventions in a structured, searchable format. By utilizing semantic embeddings and hybrid search, this skill ensures that valuable context and past decisions are always accessible to the AI, preventing knowledge silos and streamlining development workflows across multiple projects. The skill automatically handles project detection via Git metadata, ensuring that knowledge is organized correctly without manual configuration.

主な機能

01Standardized Markdown storage for version-control friendly documentation

0242 GitHub stars

03Automated recording of technical learnings and issue resolutions via dedicated scripts

04Project-specific knowledge collections with automatic Git repository detection

05Automatic index updates with auto-embedding after recording new entries

06AI-powered semantic and keyword search using qmd MCP server integration

ユースケース

01Querying historical project context to understand why specific code patterns were used

02Recording architectural decisions and 'lessons learned' during a feature implementation

03Tracking debugging steps and final resolutions for persistent or complex issues

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jellydn/my-ai-tools qmd-knowledge

For use in Claude.ai and ChatGPT

Download Skill