By jacksen-ng
Jacksen's personal closed-loop dev workflow: a project bootstrapper that stamps a local harness-engineering orchestrator + domain skills into each repo, plus scaffold / coding-standards / git-commit skills.
Professional software engineering and data science development assistant. Use when the user needs help with coding tasks, software development, data analysis, or technical implementation. Triggers include code modification requests, feature implementation, debugging, architecture decisions, dependency management, testing, or any development-related questions. Supports Chinese (Simplified), Japanese, and English.
Jacksen's standardized project structure, Docker configuration templates, and development logging conventions. Use this skill when initializing new projects, setting up development environment, or generating developer logs. Automatically applies when detecting project setup or documentation tasks.
Jacksen's git commit message style — short, single-line, conventional `<type>` prefix then a brief description. Use when writing a commit message or when the user asks to commit changes ("提交", "commit this", "帮我 commit", "write a commit message"). Covers the type set (feat / fix / docs / change / delete / deploy) and the under-50-char, imperative-mood, lowercase rules.
Bootstrap a project to use the harness workflow. Run ONCE in a repo (new or existing) when the user wants to "set up my workflow here", "初始化工作流", "scaffold CLAUDE.md and skills", or start a new project with their standard harness. It explores the repo to learn the real stack and structure, then generates THREE things into the repo — (1) a project-local `harness-engineering` orchestrator skill (the closed-loop engine, self-contained), (2) a thin CLAUDE.md (always-on rules + routing table), and (3) one domain skill per natural surface (frontend / backend / infra / data / db / …) following the proven snapshot + ground-truth-anchors template. If the repo already has skills or a CLAUDE.md, it first inventories and reconciles them against the live code — keeping what matches, proposing updates for what drifted, adding what is missing — instead of overwriting. Presents everything for review before finalizing. Complements `/dev-init` (file/folder scaffold) — this lays down the *context layer*, not the directory tree.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
My personal Claude Code workflow — a reusable closed-loop engineering harness, packaged as a plugin and distributed through this repo's own marketplace. The plugin ships a bootstrapper that stamps a self-contained workflow into any repo, so each project ends up owning its own engine — no rebuilding it by hand each time.
plugin: harness-kit (install once, available globally)
├── harness-init ← bootstrapper: explore repo → stamp the workflow into it
├── dev-init ← directory / Docker / devlog scaffold
├── coding-standards ← engineering operating principles
└── git-commit ← commit-message style
each project (generated by harness-init — self-contained, no global dependency)
├── .claude/skills/harness-engineering/ ← the closed-loop ENGINE, local to this repo
├── CLAUDE.md ← thin: hard rules + routing table
└── .claude/skills/<prefix>-<surface>/ ← domain skills, loaded on demand
Design point: the engine is per-project, not global. The plugin ships the generator
(harness-init) and the template for the engine; running it drops a tailored
harness-engineering into the repo. Every project is then self-contained — clone it and the
whole loop is right there, the same way a mature project keeps its own orchestrator.
The loop (run by each project's local harness-engineering, reusing the shared skills):
Intake (guardrails) → Plan (sprint contract) → Dispatch (load domain skill on demand) → Ground (verify against live code — map vs territory; the code wins) → Implement → Verify (separate generation from evaluation) → Gate (
/code-review+/security-review) → Report & per-action approval
/plugin marketplace add jacksen-ng/claude-workflow # register the marketplace
/plugin install harness-kit@jack-workflow # install the bootstrapper + utilities
After install, the skills are user-global (stored under ~/.claude/plugins/cache/).
dev-init — lay down the directory / Docker / devlog scaffold.harness-init — explores the repo and stamps in: a local harness-engineering orchestrator,
a thin CLAUDE.md, and one domain skill per surface — for your review.harness-engineering.Plugin installs are copied to a cache, so editing this repo does not take effect automatically:
# 1. edit a skill / template
# 2. bump "version" in plugins/harness-kit/.claude-plugin/plugin.json
# 3. push to GitHub
/plugin marketplace update jack-workflow
/reload-plugins
Note: bumping the plugin updates the bootstrapper, not orchestrators already stamped into
existing projects. Those are self-contained copies — re-run harness-init (or hand-edit) to
refresh a project's local harness-engineering.
| Path | Purpose |
|---|---|
.claude-plugin/marketplace.json | marketplace manifest (name: jack-workflow) |
plugins/harness-kit/.claude-plugin/plugin.json | plugin manifest (name: harness-kit, semver) |
plugins/harness-kit/skills/* | the 4 bundled skills (harness-init, dev-init, coding-standards, git-commit) |
plugins/harness-kit/skills/harness-init/SKILL.md | embeds Template A (engine) / B (CLAUDE.md) / C (domain skill) — the single source the bootstrapper stamps from |
npx claudepluginhub jacksen-ng/claude-workflow --plugin harness-kitReliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Intelligent draw.io diagramming plugin with AI-powered diagram generation, multi-platform embedding (GitHub, Confluence, Azure DevOps, Notion, Teams, Harness), conditional formatting, live data binding, and MCP server integration for programmatic diagram creation and management.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Design fluency for frontend development. 1 skill with 23 commands (/impeccable polish, /impeccable audit, /impeccable critique, etc.) and curated anti-pattern detection.