Technical design PRD generator. Creates an implementation-ready PRD from a technical problem definition alone, without a planning document. Triggers: "설계해", "design this", "아키텍처 PRD", "architecture PRD", "이 문제 어떻게 해결할지", "how to solve this problem", "기술적으로 어떻게", "how to technically", "PRD 만들어", "create PRD" (when no planning doc is provided). Distinction from drafti-feature: no planning document → architect, has planning document → feature.
Planning-based implementation spec PRD generator. Converts planning requirements into an implementation-ready spec. Triggers: "이 기획서로 PRD 만들어", "create PRD from this planning doc", "피쳐 PRD", "feature PRD", "기획서 기반 구현 명세", "implementation spec from planning doc", "피쳐 설계", "feature design", or when a planning document is attached/referenced. Distinction from drafti-architect: planning document exists → feature, does not exist → architect.
Decision-forcing skill. Reduces a problem to a binary fork via role decomposition (Decision / Execution / Validation / Recovery), surfaces trade-offs, forces a single choice. Triggers: "결정", "선택", "어느 쪽", "두 길", "갈피", "trade-off", "둘 중 뭐", "이전에 결정한", "다시 결정", "결정 기록", "decide", "decision", "choose between", "fork", "torn between", "past decision", "decided before", "record decision". Scope: any domain. Pre-PRD, pre-implementation.
Autonomous implementation engine. PRD to task decomposition, RALP loop autonomous execution, cross-session context preservation, experience accumulation. triggers: "구현 시작", "start implementation", "태스크 분해", "decompose tasks", "루프 돌려", "run loop", "이어서 진행", "continue", "다음 태스크", "next task", "진행 상태", "progress status", "자산 현황", "asset status", "자산 압축", "compress assets", "이 패턴 기억해", "remember this pattern", "스킬로 만들어", "make it a skill", Request to resume work when harnish-current-work.json exists.
Inspection skill. Triggers: "점검해", "확인해", "검증해", "ralphi", "셀프점검", "커버리지 확인", "테스트 갭", "고쳐", "수정해", "점검하고 고쳐", "자동으로 처리해", "inspect", "check", "verify", "self-check", "coverage check", "test gap", "fix", "repair", "inspect and fix", "handle automatically"
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
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.
Claude Code plugin — autonomous implementation engine
harnish (harness + ish) — an implementation environment that gets smarter as you work. Failures become guardrails, patterns accumulate, and context persists across sessions.
| Skill | Version | Command | Role |
|---|---|---|---|
| forki | 0.0.1 | /harnish:forki | Decision forcing (binary fork + D/E/V/R + trade-off, HITL only) |
| drafti-architect | 0.0.1 | /harnish:drafti-architect | Tech-driven design PRD generation |
| drafti-feature | 0.0.1 | /harnish:drafti-feature | Planning-based implementation spec PRD |
| harnish | 0.0.1 | /harnish:harnish | Autonomous implementation engine (seeding + RALP loop + anchoring + experience) |
| ralphi | 0.0.1 | /harnish:ralphi | Inspection (HITL reporting or autonomous fix) |
Each skill operates in an independent orbit, connected only through shared artifacts (files).
forki ──→ forces a binary decision (D/E/V/R + trade-off, HITL only)
↓
drafti ──→ docs/prd-*.md ──→ harnish ──→ implementation code
│
└── .harnish/ (work coordinates + experience, in user project CWD)
ralphi ──→ inspects any artifact (PRD, SKILL.md, scripts, code)
HITL (report → wait) or autonomous (fix immediately)
User: "Should we use Postgres or MongoDB for this?"
→ forki frames as binary → asks user to confirm A/B
→ asks user to fill the D/E/V/R table (8 cells)
→ surfaces trade-off → asks user to commit (LLM cannot decide)
→ outputs the structural reason for the choice
User: "Design a Redis cache layer"
→ drafti-architect explores 2-3 design alternatives with trade-off analysis
→ generates docs/prd-redis-cache.md
User: "Create a PRD from this planning doc" (with planning document attached)
→ drafti-feature generates implementation spec PRD (feature flags only when needed)
→ generates docs/prd-user-profile-edit.md
User: "Start implementation" or "Decompose tasks"
→ Decomposes PRD into atomic tasks → generates harnish-current-work.json
→ "3 Phases, 12 Tasks seeded — review then 'run the loop'"
User: "Run the loop"
→ RALP loop auto-executes (Read → Act → Log → Progress → repeat)
→ Updates harnish-current-work.json every 3 actions, milestone report on phase completion
User: (in a new session) "Continue where I left off"
→ Restores coordinates from harnish-current-work.json, auto-resumes from break point
User: "Inspect this PRD"
→ Type detection (PRD) → static analysis → issue report → waits for user judgment (HITL)
User: "Inspect and fix src/cache.py"
→ Type detection (code) → analysis → immediate fix → result report (autonomous)
→ Rolls back on test failure, classifies as unfixed when intent is unclear
User: "Remember this pattern"
→ Records as pattern asset → auto-referenced in future work
User: "Asset status"
→ Shows accumulated failure/pattern/guardrail/snippet/decision assets
User: "Make this a skill"
→ Generates reusable SKILL.md draft from compressed assets
harnish/
├── .claude-plugin/plugin.json # Plugin manifest
├── skills/
│ ├── forki/ # Decision forcing (binary fork + D/E/V/R + trade-off, HITL only)
│ ├── drafti-architect/ # Tech design PRD generation
│ ├── drafti-feature/ # Planning spec PRD generation
│ ├── harnish/ # Autonomous implementation (seeding/RALP/anchoring/experience)
│ └── ralphi/ # Inspection (HITL/autonomous)
├── hooks/hooks.json # Claude Code hooks
├── scripts/ # Shared scripts (16)
├── docs/ # PRD documents
├── VERSION # Repo version
├── CHANGELOG.md # Release history
└── VERSIONING.md # Versioning policy
/plugin marketplace add jazz1x/harnish
/plugin install harnish@harnish
For GitLab or other self-hosted git services:
/plugin marketplace add https://gitlab.com/your-org/harnish.git
/plugin install harnish@harnish
git clone https://github.com/jazz1x/harnish.git
cd your-project
claude --plugin-dir /path/to/harnish
Skills register as /harnish:forki, /harnish:harnish, /harnish:drafti-architect, /harnish:drafti-feature, /harnish:ralphi.
Three ways to use this repo as a base:
Copy one skill directly into your own project — no plugin install needed.
mkdir -p .claude/skills
cp -r /path/to/harnish/skills/forki .claude/skills/
npx claudepluginhub jhu900202/plugins --plugin harnishSpec-driven development for big features. When features get too big, plan mode gets too vague—leading to hallucinations during implementation. ShipSpec replaces vague plans with structured PRDs, technical designs, and ordered tasks that keep Claude grounded.
Provider-agnostic skills for autonomous product development: spec, task, implement, test, review, and summarize changes.
AI Coding Agent - PRD to tasks to implementation loop
Autonomous development methodology: PRD interviews → agent execution → automated review
AI-powered cascading development framework with design document system and multi-agent collaboration. Breaks down projects into Features (Mega Plan), Features into Stories (Hybrid Ralph), with auto-generated technical design docs, dependency-driven batch execution, Git Worktree isolation, and support for multiple AI agents (Codex, Amp, Aider, etc.).
Specification-driven development workflow: specify → plan → tasks → implement