junjak-ai-harness

Plugin junjak-ai-harness from marketplace ai-harness in JunjaK/ai-harness. Commands read from the repo's manifest — 5 days ago.

Skills (20)

• agentic-testing

  On-demand, adapter-based agent QA. Verify archived /team-qa scenarios or a directly supplied outcome through a real UI/API driver; record passed, failed, or blocked evidence. Test generation is opt-in via --crystallize.

• brain-connect

  Wire an optional personal 'brain' SSOT (a separate repo holding your cross-machine persona, global CLAUDE.md, personal global skills, harness auto-memory, and a recommended-settings manifest) into the Claude Code environment on a machine. Use when setting up the harness on a new machine (work/home),

• brainstorm

  Use BEFORE any solo/low-blast creative work — designing a feature, component, behavior change, or this harness itself — to turn an idea into an approved design through a lightweight collaborative dialogue. The solo counterpart to /team-brainstorm (which owns multi-file / cross-cutting / product / ar

• browser-e2e

  The only browser driver (ego lite). Use it for any task that drives a real browser, named or not: E2E, QA, smoke, exploration, login-gated checks, live selectors, team verification, /team-qa. No ego-browser → show the install steps and stop; never switch to another browser tool. Committed .spec.ts s

• checkpoint

  Save and restore work state across sessions, branches, and compactions. Use when saving progress before stopping, switching branches, or when the /checkpoint command is invoked.

• continuous-learning

  Extract reusable patterns from sessions, manage session state, and evolve learnings into skills. Use at session end, after completing features, or when patterns emerge during work.

• contract-sync

  Regenerate and re-verify a code-generated API client after any backend/server contract change, BEFORE writing or verifying client code against it. Use whenever a task changes a backend DB schema, request/response DTO, enum, or endpoint shape in a project where the frontend consumes a generated clien

• debug

  LSP-driven debugging patterns for TypeScript codebases under the harness root-cause-first method (Iron Law: no fix before the root cause is confirmed). Use in Phase 3-4 escalation scenarios or when investigating unexpected behavior; `/debug` is the solo entry point with the full method and red flags

• docs-lifecycle

  Manage the lifecycle of project documents (intent, plans, specs, findings) in `_docs/` — the lifecycle-vs-collection bucket split, status↔folder lockstep, date/topic foldering, reference-safe moves, and merging a task's scattered sidecar docs into ONE archive on completion. Use whenever an intent/pl

• greenfield-bootstrap

  Bootstrap a brand-new project from zero — research the stack, scaffold it, and generate the project profile — when there is no code yet. Use when /team-new is invoked, or when the user wants to start a new project from an idea. Ends by handing off to /team-run for the first feature. Distinct from pr

• handoff

  Write a handoff document — a state layer over a spec/plan capturing what's done, what's left, and how to resume — when work passes to another agent or a future session. Use at session end with unfinished work, before a compaction that ends active work, or when explicitly handing a task off. Writes i

• parallelization

  Git worktree management and instance scaling guidelines. Use when planning parallel agent work, managing worktrees, or deciding when to scale instances.

• plan-visualizer

  Generate an interactive HTML visualization of a team plan — team composition, phase flow, file assignments, dependency graph — saved alongside the plan .md. Fires ONLY from the /plan-visualizer command, i.e. when the user explicitly asks for a diagram. No workflow phase invokes it.

• project-analyzer

  Analyze project structure, conventions, and patterns to generate a project profile. Use when /team-init is invoked or when team agents need to understand the codebase. Scans package.json, config files, source code patterns, and test setup to produce structured profiles in .claude/project-profile/.

• submodule-worktree

  Worktree strategy for a submodule-monorepo where the superproject is a thin shell (doc storage + submodule pointers) and the real code lives in submodules (illustratively `fe`/`be`). TRIGGER on ONE condition: a worktree / parallel-implementation request lands in a project whose profile (`.claude/pro

• take-over

  Take over a handed-off work-stream — the read/resume counterpart to the `handoff` skill. A handoff is a temp carrier: locate the `_docs/handoff/` doc, hydrate its linked spec (follow `related:`), verify its state claims against the actual repo (commits, branch, tests), then GRADUATE it into its dura

• team-workflow

  Multi-agent team workflow orchestration with 5 phases and TDD enforcement. Use when /team or /team-run commands are invoked. Coordinates planning, parallel implementation, a lightweight merged-tree verification pass, final security review, and deferred QA item capture. Run accumulated QA later with

• test-scenario-doc

  Generate a self-contained, interactive HTML test-scenario / QA checklist document — grouped scenarios with checkable steps, pass/fail/hold verdicts, a sticky progress bar, localStorage auto-save, a copyable markdown summary, an image-embedded HTML result report, and per-scenario screenshot attachmen

• verify

  Run the project's authoritative gates from project-profile verify.json and produce a 됨 / 미검증 / 안 됨 evidence block — before reporting any change done or fixed, in team Phase 4, and after a /debug fix. Also plants a violation to prove a new check catches it.

• wiki

  Build and maintain the agent wiki (.claude/wiki/) — a persistent, compounding knowledge base. Use when ingesting a source into the wiki, querying it, or linting it for drift. The wiki links to the SSOT (code / _docs/); it routes and synthesizes, it does NOT duplicate facts.

Slash commands (15)

• /junjak-ai-harness:brain-connect

  Brain Connect — pair an optional personal brain SSOT (cross-machine persona, global CLAUDE.md, personal global skills, auto-memory, recommended-settings manifest) with the harness on a machine, or relocate an existing one

• /junjak-ai-harness:checkpoint

  Checkpoint — save or restore work state across sessions, branches, and compactions; the harness work-state command, distinct from Claude Code's built-in /resume.

• /junjak-ai-harness:debug

  Debug — solo systematic debug of a bug / test failure / unexpected behavior under the Iron Law (root cause before any fix); layers the TS/LSP debug skill for TypeScript targets, escalates to /team when it turns Fundamental. The non-team-workflow debugging entry point.

• /junjak-ai-harness:docs-sweep

  Docs Sweep — drain stale active docs (reap) + enforce the 6 orphan-mode invariants (lint) in _docs/

• /junjak-ai-harness:meta-prompt

  Meta-Prompt — compile a raw context dump into an optimized, self-contained prompt to inject into a fresh environment (new session, /team-run task string, subagent, or another tool). Borrows brainstorm's questioning discipline; terminal output is a portable prompt, not a design doc.

• /junjak-ai-harness:plan-visualizer

  Plan Visualizer — render a finalized team plan as an interactive HTML diagram (explicit request only; no phase auto-fires it)

• /junjak-ai-harness:take-over

  Take Over — resume a handed-off work-stream: hydrate the spec, verify the handoff's state against the repo, then graduate the handoff into its durable _docs home (complete/ or a plan)

• /junjak-ai-harness:team-brainstorm

  Team Brainstorm — Leader + Architects가 plan만 논의 (구현 없음, 시각화 포함)

• /junjak-ai-harness:team-init

  Team Init — 프로젝트 구조 분석 및 프로필 생성/갱신

• /junjak-ai-harness:team-new

  Team New — greenfield project bootstrap: deep-research → architect stack decision → user gate → scaffold → profile

• /junjak-ai-harness:team-qa

  Run deferred team QA on demand from completed plan archives; verify a bounded batch and record evidence without restarting implementation.

• /junjak-ai-harness:team-run

  Team Workflow (Autonomous) — multi-agent team workflow with full auto-execution

• /junjak-ai-harness:team

  Team Workflow (Interactive) — multi-agent team workflow with user involvement in planning

• /junjak-ai-harness:test-scenario-doc

  Test Scenario Doc — generate an interactive human QA checklist HTML on demand (human-requested manual verification)

• /junjak-ai-harness:worktree-deps

  Worktree Deps — provision a worktree's node_modules via the package manager's shared content-addressable store (hard-link, no re-download, parallel-safe). pnpm/bun handled inline; any other manager resolved via Context7. Never symlinks node_modules across worktrees.

Subagents (10)

• team-agentic-tester

  Standalone deferred QA executor for /team-qa. Verifies selected archived scenarios and reports evidence; generates regression tests only when explicitly requested.

• team-architect-be

  Backend architecture specialist — detailed backend/API plans for team workflow

• team-architect-fe

  Frontend architecture specialist — detailed frontend plans for team workflow

• team-architect-infra

  Infrastructure and security specialist — on-demand review for team workflow (Phase 1 + Phase 5)

• team-designer

  TDD-based code developer — implements features following Red-Green-Refactor cycle in isolated worktrees

• team-leader

  Team workflow leader — drafts plans, coordinates architects, manages approval gates and escalation

• team-tester

  Focused verification specialist for the team workflow

• team-uiux-master

  Senior UI/UX specialist — reviews plans for visual changes, proposes high-quality UI/UX modifications with design intelligence, enforces accessibility and interaction standards

• web-architect

  Web frontend/fullstack architecture specialist — designs component structures, state management, API layers, routing, performance strategies, and bundle optimization for web development requests.

• web-reviewer

  Web quality reviewer — validates accessibility (WCAG 2.2), performance (Core Web Vitals), SEO, design quality, and AI Slop detection for web implementations.

MCP servers (1)

• jev

  runs: node ${CLAUDE_PLUGIN_ROOT}/mcp/jev.mjs

Hooks (11)

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/session-start.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/session-compact.sh"

• Stop

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/session-stop.sh"

• Stop

  node "${CLAUDE_PLUGIN_ROOT}/hooks/reply-language.mjs"

• Stop
• SessionEnd

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/session-end.sh"

• PreCompact

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/pre-compact.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/guardrails.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/browser-guard.sh"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/browser-guard.sh"

• PostToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/post-edit-warn.sh"

Similar plugins