A growing collection of Claude Code skills for developers.
From GitHub (once the repo is public):
/plugins marketplace add github:kumarishan/kistack
/plugins install kistack
From a local clone:
/plugins marketplace add /path/to/kistack
/plugins install kistack
Then restart Claude Code for skills to activate.
| Skill | What it does |
|---|---|
repo-tutorial |
Turn any code repository into a hands-on, chapter-by-chapter tutorial |
ade |
ADE lead skill — coordinate end-to-end feature development across repos, from PRD to merged PR |
ade-prd-review |
Clarify and refine a PRD with structured questions and testable acceptance criteria |
ade-techspec-high |
Generate a cross-repo high-level technical spec from a PRD |
ade-techspec-low |
Generate per-repo low-level implementation specs from a high-level spec |
ade-agent-plan |
Decompose a low-level spec into a DAG of discrete agent execution tasks |
ade-test-scenarios |
Generate integration test (IT) and E2E test scenarios from specs |
ade-execute |
Implement code changes in git worktrees following the agent plan |
ade-uat |
Validate the implementation against test scenarios and PRD acceptance criteria |
ade-pr |
Commit, push, and create GitHub PRs for all repos using gh |
ade-pr-feedback |
Fetch PR review comments and incorporate them into the worktrees |
ade-close |
Remove worktrees, delete branches, and archive the project after merge |
ADE is a suite of coordinated skills for end-to-end feature development. Start with:
ade new
ADE stores project state in ~/.ade/<proj-name>/config.json and creates git worktrees for each repo so you can work across multiple terminal sessions without disrupting your main branches.
Typical flow:
ade new → initialize project, create worktrees
[provide PRD] → prd-review, high-level spec, low-level spec
ade agent-plan → generate task DAG + test scenarios
ade execute → implement code in worktrees
ade uat → validate against test scenarios
ade pr → create GitHub PRs
ade pr-feedback → incorporate review comments
ade close → clean up after merge
Skills activate automatically when relevant. You can also invoke them explicitly:
create a tutorial for this repo
teach me this codebase
build a learning guide for the auth module
Each skill lives in skills/<skill-name>/SKILL.md. To add a new skill:
- Create
skills/your-skill-name/SKILL.md - Add frontmatter with
nameanddescription - Write the skill body (see existing skills for reference)
- Update the skills table in this README
- Bump the version in
.claude-plugin/plugin.json
MIT