
wwyharry/skillforge-core
Developer toolsこのページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る
SkillForge is a local-first repository distillation workbench.
概要
SkillForge is a repository distillation workbench that turns messy project materials into reusable AI skill packages. It scans local folders, parses mixed document formats, extracts workflow evidence, clusters that evidence into capabilities, and compiles the result into previewable and exportable skill outputs through a built-in web UI. Teams already have the raw knowledge they need: product docs, design notes, research reports, runbooks, SOPs, meeting notes, API references, and random markdown files scattered across repositories and shared folders. The hard part is turning that fragmented, implicit knowledge into something reusable by an AI agent.
README
以下は原文 README の抜粋です。原文を見る。
SkillForge
English | 简体中文
SkillForge is a repository distillation workbench that turns messy project materials into reusable AI skill packages.
It scans local folders, parses mixed document formats, extracts workflow evidence, clusters that evidence into capabilities, and compiles the result into previewable and exportable skill outputs through a built-in web UI.
Why SkillForge?
Teams already have the raw knowledge they need: product docs, design notes, research reports, runbooks, SOPs, meeting notes, API references, and random markdown files scattered across repositories and shared folders.
The hard part is turning that fragmented, implicit knowledge into something reusable by an AI agent.
Doing that manually is slow and brittle:
- one-off scripts are fast to start but hard to maintain
- generic agent frameworks help orchestration, but they do not automatically distill repository knowledge into skill packages
- manual copy/paste into prompts does not scale across teams or projects
SkillForge focuses on that missing step:
evidence extraction → capability clustering → skill compilation
So instead of just “running an agent,” you can systematically convert repository knowledge into reusable skill artifacts.
What makes it different?
Compared with ad-hoc scripts
Ad-hoc scripts are useful for a single extraction pass, but they usually stop at file parsing or keyword search.
SkillForge gives you:
- a repeatable end-to-end pipeline
- structured job tracking
- a visual workflow dashboard
- preview and export for generated skill outputs
Compared with LangChain / DSPy / custom agent stacks
Those tools are great for orchestration, prompting, and model workflows. SkillForge solves a narrower but very practical problem: turning a real-world repository into reusable skill packages with a local-first review workflow.
In short:
- LangChain / DSPy help you build agent systems
- SkillForge helps you distill repository knowledge into agent-usable skills
They are complementary, not mutually exclusive.
Core use cases
- Distill internal documentation into reusable team skills
- Convert research folders into structured analyst workflows
- Extract SOPs and process knowledge from operations repositories
- Turn API docs and implementation notes into integration-oriented skills
- Build a reviewable bridge between messy source materials and AI-ready skill packages
Product walkthrough
SkillForge is designed as a visual, local-first workflow.
Visual workflow
推奨ツール
別のキーワードを試すか、フィルタを外してください。
インストール
npx skillfish add wwyharry/skillforge-core