AgentDojo

56 AI skills across 5 domains for Claude Code — Product Management, Principal Engineering, Senior SDE, Quality Assurance, and UI/UX Design.

One person. Entire engineering org.

/sde-tdd my-feature          # Write code test-first
/gap-analyst                 # Find product gaps competitors have
/ux-design-system            # Build a full design system
/qae-security                # OWASP Top 10 security scan
/pe-tech-strategy scaling    # Write engineering strategy
/orchestrator                # Run the full 9-phase SDLC

---

What Is This?

AgentDojo is a skill system for Claude Code. Each skill is a standalone agent — type a slash command and it reads your codebase, analyzes the problem, and produces real output (code, docs, designs, test plans).

Not a framework. Not a library. Just skills you install and use.

---

Quick Start

# Clone
git clone https://github.com/mission-agi/AgentDojo.git

# Copy slash commands to Claude
cp -r AgentDojo/claude/commands/ ~/.claude/commands/

# Copy orchestrator
mkdir -p ~/.claude/skills/orchestrator
cp AgentDojo/orchestrator/SKILL.md ~/.claude/skills/orchestrator/SKILL.md

# Copy domain agents
mkdir -p ~/.claude/skills/agents
cp AgentDojo/agents/*.md ~/.claude/skills/agents/

# Done. Open Claude Code and type any skill command.

Verify It Works

claude
# Then type:
/sde-tdd my-feature

If the skill loads, you're good.

Requirements

• Claude Code CLI
• A project repository to work in

---

All 60 Commands

Product Management (11 skills)

Command	What It Does
/gap-analyst	Find product gaps, deduplicate against backlog, score with WINNING filter
/prd-generator [feature]	Generate structured PRDs, file as GitHub Issues
/research-agent [company]	Deep competitor profiling — features, pricing, sentiment
/discovery-validator [idea]	Turn ideas into testable hypotheses
/customer-research [topic]	Interview scripts, personas, jobs-to-be-done
/experiment-designer [experiment]	A/B test plans, hypothesis cards, cohort analysis
/prioritization-engine	RICE/Kano/MoSCoW/ICE scoring, outcome-based roadmaps
/stakeholder-communicator [type]	Status reports, stakeholder maps, decision logs
/metrics-advisor	North star metrics, metric trees, OKR templates
/pivot-analyzer	PMF signal assessment, pivot type mapping
/buyer-psychology [segment]	Buyer profiles, cognitive biases, switching forces

Principal Engineering (11 skills)

Command	What It Does
/tech-strategy [topic]	Engineering strategy & vision, strategy kernel
/architecture-reviewer [RFC]	RFC review, ADR writing, design quality assessment
/technical-quality [system]	7-tier quality progression, DORA metrics, tech debt
/migration-planner [migration]	De-Risk/Enable/Finish framework for migrations
/leverage-analyzer [situation]	Impact/time analysis, anti-pattern detection
/org-health [concern]	Team state diagnosis, chain-link bottleneck analysis
/influence-navigator [situation]	Nemawashi, radiating intent, stakeholder alignment
/decision-facilitator [topic]	Decision classification, facilitation templates
/mentorship-sponsorship [person]	Growth plans, sponsorship playbook, feedback
/incident-reliability [incident]	Post-mortems, reliability reviews, on-call optimization
/career-navigator [goal]	Staff+ archetypes, promotion packets

Senior SDE (11 skills)

Command	What It Does
/sde-tdd [feature]	Red-green-refactor, testing pyramid, test doubles
/sde-code-craftsman [topic]	Clean code, DRY, naming, guard clauses
/sde-system-design [system]	4-step framework, CAP theorem, trade-off analysis
/sde-architecture [topic]	SOLID principles, clean architecture layers
/sde-data-systems [topic]	Data models, replication, partitioning, consistency
/sde-requirements [feature]	Elicitation, specification, traceability matrices
/sde-systems-thinking [system]	Feedback loops, leverage points, system archetypes
/sde-ml-design [system]	ML pipelines, feature stores, model serving
/sde-debugging [bug]	Systematic debugging, root cause analysis
/sde-estimation [project]	PERT estimation, Brooks's Law, risk buffers
/sde-code-review [PR]	Review checklists, SOLID compliance, security

Quality Assurance (11 skills)

Command	What It Does
/qae-test-strategy [project]	Risk-based test strategies, coverage models
/qae-test-plan [release]	Test plans with entry/exit criteria
/qae-exploratory [feature]	Session-based exploratory testing, charters
/qae-cicd-pipeline [project]	Pipeline design, quality gates, deployment strategies
/qae-automation [project]	Framework selection, automation pyramid, ROI
/qae-api-testing [API]	REST/GraphQL testing, contract testing
/qae-performance [system]	Load/stress/soak testing, capacity planning
/qae-security [system]	OWASP Top 10, STRIDE threat modeling
/qae-test-data [project]	Data generation, masking, synthetic data
/qae-defect-management [project]	Bug lifecycle, root cause analysis (5 Whys)
/qae-quality-metrics [release]	Coverage metrics, defect density, release readiness

UI/UX Design (11 skills)

Command	What It Does
/ux-design-system [project]	Atomic Design, design tokens, component maturity
/ux-visual-hierarchy [page]	8pt grid, spacing scales, Gestalt principles
/ux-typography [project]	Type scales, font pairing, clamp() fluid sizing
/ux-color-system [project]	HSL palettes, WCAG contrast, dark mode tokens
/ux-component-design [component]	State matrices, all states + sizes
/ux-landing-page [product]	Hero patterns, CTA hierarchy, conversion optimization
/ux-interaction-design [component]	Micro-interactions, CSS timing, motion systems
/ux-responsive [page]	Mobile-first CSS, breakpoints, container queries
/ux-dashboard [dashboard]	Chart selection, KPI cards, data tables
/ux-accessibility [page]	WCAG 2.2 compliance, keyboard nav, ARIA
/ux-review [design]	Nielsen's 10 Heuristics, scoring rubrics

Orchestrators (5 commands)

Command	What It Does
/pm-orchestrator [goal]	Chain all 11 PM skills in sequence
/pe-orchestrator [goal]	Chain all 11 PE skills in sequence
/sde-orchestrator [goal]	Chain all 11 SDE skills in sequence
/qae-orchestrator [goal]	Chain all 11 QAE skills in sequence
/ux-orchestrator [goal]	Chain all 11 UX skills in sequence
/orchestrator [feature]	Master Orchestrator — 9-phase SDLC across all domains

---

How It Works

Skills Read Your Code

Every skill reads your actual codebase. /sde-code-review reads your PR diffs. /qae-security scans your source files. /ux-accessibility audits your HTML. No dummy output — real analysis of real code.

Skills Write Real Output

Skills produce files in domain-specific directories:

pm/    — PRDs, gap analyses, competitor profiles, experiment briefs
pe/    — Strategy docs, ADRs, quality assessments, migration plans
sde/   — Architecture reviews, TDD sessions, system designs
qae/   — Test strategies, CI/CD configs, security reports
ux/    — Design tokens, component specs, accessibility audits

Skills Chain Together

Output from one skill feeds the next:

Research Agent → Gap Analyst → PRD Generator → GitHub Issues
                                    ↓
                        UX Design System → Component Design
                                    ↓
                           SDE TDD → Code Review
                                    ↓
                        QAE Test Strategy → CI/CD Pipeline

Master Orchestrator

The /orchestrator command runs the full product lifecycle in 9 phases:

Phase 1: Discovery        — PM research, gap analysis
Phase 2: Strategy          — PE tech strategy, UX design system
Phase 3: Architecture      — PE architecture review
Phase 4: Requirements      — PM PRD, SDE requirements
Phase 5: Design            — UX components, SDE system design
Phase 6: Implementation    — SDE TDD, code craftsman (with quality gates)
Phase 7: Quality           — QAE test strategy, security, performance
Phase 8: Integration       — Cross-domain validation
Phase 9: Launch            — Metrics, monitoring, retrospective

4 user decision gates at strategic boundaries. Automatic skill loop exit criteria. Cross-domain feedback loops (SDE finds bugs → QAE validates → SDE fixes).

---

Alternative Installation

As a Claude Plugin

git clone https://github.com/mission-agi/AgentDojo.git
mkdir -p ~/.claude/plugins/cache/local/agentdojo/1.0.0
cp -r AgentDojo/* ~/.claude/plugins/cache/local/agentdojo/1.0.0/

Directly in a Project

git clone https://github.com/mission-agi/AgentDojo.git .agentdojo
cp -r .agentdojo/claude/commands/ .claude/commands/

---

Project Structure

AgentDojo/
├── CLAUDE.md              # Claude Code project instructions (auto-loaded)
├── README.md              # This file
├── LICENSE                # All Rights Reserved
│
├── claude/commands/       # Slash commands (the skills you type)
│   ├── sde-tdd.md         #   /sde-tdd
│   ├── gap-analyst.md     #   /gap-analyst
│   ├── ux-design-system.md#   /ux-design-system
│   └── ... (60 files)     #   one file per skill + orchestrator
│
├── agents/                # Domain agent definitions
│   ├── pm.md              #   PM agent — routes to 11 PM skills
│   ├── pe.md              #   PE agent — routes to 11 PE skills
│   ├── sde.md             #   SDE agent — routes to 11 SDE skills
│   ├── qae.md             #   QAE agent — routes to 11 QAE skills
│   ├── ux.md              #   UX agent — routes to 11 UX skills
│   └── orchestrator.md    #   Master orchestrator agent
│
├── orchestrator/          # Master orchestrator engine
│   ├── SKILL.md           #   9-phase SDLC logic, gates, skill selection
│   ├── routing-rules.json #   Phase transitions, loop exit criteria
│   └── coordination-service.md  # Cross-domain handoff protocols
│
├── pm/                    # Product Management skill definitions
├── pe/                    # Principal Engineering skill definitions
├── sde/                   # Senior SDE skill definitions
├── qae/                   # Quality Assurance skill definitions
└── ux/                    # UI/UX Design skill definitions

What Each Part Does

Folder/File	What It Is	When You Need It	How To Use It
claude/commands/	The slash commands themselves	Always — this is the core	Copy to ~/.claude/commands/ and type /sde-tdd in Claude Code
CLAUDE.md	Project context Claude reads automatically	Always — tells Claude the conventions	Lives in project root, Claude loads it on startup
agents/	Agent definitions for multi-skill orchestration	When using orchestrators (/pm-orchestrator, /orchestrator)	Copy to ~/.claude/skills/agents/
orchestrator/	Master orchestrator engine	When running the full 9-phase SDLC (/orchestrator)	Copy to ~/.claude/skills/orchestrator/
pm/ pe/ sde/ qae/ ux/	Skill definitions + example outputs	Reference — see how each skill is structured	Skills are read by the commands in claude/commands/
project/	Skill manifest	Internal routing metadata	Used by the orchestrator for skill dependency tracking

What You Actually Install

For most users, you only need to copy 3 things:

# 1. Slash commands (required)
cp -r claude/commands/ ~/.claude/commands/

# 2. Orchestrator skill (if you want /orchestrator)
mkdir -p ~/.claude/skills/orchestrator
cp orchestrator/SKILL.md ~/.claude/skills/orchestrator/SKILL.md

# 3. Domain agents (if you want /pm-orchestrator, /sde-orchestrator, etc.)
mkdir -p ~/.claude/skills/agents
cp agents/*.md ~/.claude/skills/agents/

If you only want individual skills (like /sde-tdd or /gap-analyst), step 1 alone is enough.

---

License

All Rights Reserved. See LICENSE.