Enterprise-grade development workflow for Claude Code
Traceable. Controllable. Auditable.
AI-assisted development is fast — but opaque.
Decisions vanish with the conversation. Stakeholders can't review what was decided or why.
This plugin turns that around.
| Multi-phase investigation, PoC validation, and structured test planning — before any code is written | Document per phase, citation-aware research, task-based git branching — every decision recorded | Approval gates at every phase, structured decision records — correct course before work compounds |
In Claude Code:
/plugin marketplace add arudita-zzz/enterprise-dev-suite
/plugin install enterprise-dev-suite@enterprise-dev-suite
Then start your first workflow:
/enterprise-dev-suite:feasibility-study
flowchart LR
D["/deep-research"] -. "research_report.md\n(-r flag)" .-> F["/feasibility-study"]
F -- "feasibility_report.md" --> S["/solution-design"]
S -- "solution_design.md" --> I["/implement-tdd"]
I -- "implementation_report.md" --> G["/generate-slides"]
style D fill:#f3e5f5,stroke:#9C27B0
style F fill:#e8f4fd,stroke:#2196F3
style S fill:#fff3e0,stroke:#FF9800
style I fill:#e8f5e9,stroke:#4CAF50
style G fill:#fce4ec,stroke:#E91E63
| Phase | What happens | Output | Gate |
|---|---|---|---|
| Deep Research (optional) | Autonomous multi-task investigation with adaptive re-evaluation | research_report.md |
Topic approval |
| Feasibility Study | Codebase analysis, web research, solution candidates, optional PoC | feasibility_report.md |
Approach selection |
| Solution Design | Subtask breakdown, dependency mapping, test case planning | solution_design.md |
Subtask approval |
| TDD Implementation | Test-driven development with per-subtask quality gates | implementation_report.md |
Per-subtask design approval |
| Slide Generation | Marp presentation from phase artifacts for technical review | slides.md |
— |
1. Write a feature spec at .claude/claudeRes/scripts/feature_spec.md — describe what you want to build.
2. Run the workflow:
/enterprise-dev-suite:feasibility-study ← analyzes codebase, proposes approaches
/enterprise-dev-suite:solution-design ← breaks into subtasks with test cases
/enterprise-dev-suite:implement-tdd ← implements via TDD, subtask by subtask
3. Review artifacts — each phase produces a structured document in your docs directory.
That's it. Every decision is recorded, every phase is reviewable.
/enterprise-dev-suite:deep-research "your research topic"
→ Approve the task plan, then wait. The investigation runs fully autonomously.
Feed the results into the development workflow:
/enterprise-dev-suite:feasibility-study -r {path-to-research_report.md}
- Autonomous task decomposition — breaks your topic into 3-8 investigation tasks automatically
- Adaptive re-evaluation — dynamically revises remaining tasks after each completion
- Multiple investigation methods — web-research, codebase-analysis, comparative, synthesis
- Gap analysis & auto-fill — detects coverage gaps and spawns additional tasks to close them
- Fully autonomous after approval — approve once, get a comprehensive report back
| Plan Mode | enterprise-dev-suite | |
|---|---|---|
| Workflow | Ad-hoc each time | Pre-built 3-phase with user approval gates |
| Artifacts | Lost with conversation | Persisted files, resumable across sessions |
| Decision records | Not retained | Built into feasibility report |
| Correctability | None | User approval at every key decision point |
| Output consistency | Varies per run | Unified structure via templates |
| Specialized agents | General-purpose only | 5 dedicated agents + autonomous deep research |
| Audit trail | None | Full document chain with citations |
| Resumability | Lost on session end | File artifacts, cross-session |
Best fit: Teams working on large codebases where development decisions need stakeholder review.
| Skill | Description |
|---|---|
deep-research |
Autonomous multi-task investigation with adaptive re-evaluation |
feasibility-study |
Codebase analysis, web research, solution candidates, PoC |
solution-design |
Subtask breakdown, test cases, precedence diagram |
implement-tdd |
TDD implementation per subtask |
small-feature |
All-in-one quick implementation with TDD |
generate-slides |
Marp presentation slides from phase artifacts |
| Agent | Purpose |
|---|---|
codebase-investigator |
3-phase iterative codebase exploration |
poc-feasibility-expert |
PoC implementation and validation |
tdd-implementer |
TDD-based implementation |
web-research-expert |
Technical research and documentation |
document-summarizer |
Document summary generation |
Works out-of-the-box. Customize defaults (docs directory, language, etc.) by editing conventions.md in the plugin directory.
Register project-specific investigation tools via the Custom Investigation Tools table in conventions.md — they're automatically invoked during investigation phases.
See plugin README for full documentation.
enterprise-dev-suite/
├── plugins/
│ └── enterprise-dev-suite/ # Main plugin
│ ├── skills/ # 6 user-invocable skills
│ ├── agents/ # 5 specialized subagents
│ ├── script/ # Shell scripts for tooling
│ ├── doc/ # Workflow diagrams
│ ├── conventions.md # Configurable defaults
│ └── README.md # Detailed documentation
└── README.md
# Optional: Deep research on a topic
/enterprise-dev-suite:deep-research "MCP server integration patterns"
# → research_report.md
# Phase 1: Feasibility Study (with prior research)
/enterprise-dev-suite:feasibility-study -r {path-to-research_report.md}
# → feasibility/feasibility_report.md
# Phase 2: Solution Design
/enterprise-dev-suite:solution-design
# → design/solution_design.md
# Phase 3: TDD Implementation
/enterprise-dev-suite:implement-tdd
# → implementation/implementation_report.md
# Generate presentation for review
/enterprise-dev-suite:generate-slides
# → slides_{task_name}.mdContributions welcome! See CONTRIBUTING.md for guidelines.
MIT