ワンクリックで
draft
Conversational drafting for a specific concern area when documentation doesn't exist. Ask targeted questions and generate draft SCDs.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Conversational drafting for a specific concern area when documentation doesn't exist. Ask targeted questions and generate draft SCDs.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Process a document (PRD, architecture doc, security requirements, etc.) into structured context. Extracts relevant information and generates appropriate SCDs.
Initialize structured context for this project. Scaffolds .scs/ with all 11 concern bundles, scans the project, and recommends where to start.
Show what structured context exists and what's missing. Overview of bundles, SCDs, coverage across all 11 concerns, and compilation sync status.
Add known compliance or regulatory standards (HIPAA, SOC2, PCI, CHAI, GDPR) to your project context. Copies pre-built standards SCDs from the plugin's standards library.
Check structure and surface issues for human review. Validates YAML, references, concern coverage, and compilation status.
Lock bundles with semantic versioning. Guides through version selection, updates files, and integrates with git.
| name | draft |
| description | Conversational drafting for a specific concern area when documentation doesn't exist. Ask targeted questions and generate draft SCDs. |
| argument-hint | <concern: architecture|security|performance|usability|compliance|data|testing|deployment|safety|ethics|business> |
| allowed-tools | Read, Glob, Grep, Write, Bash(mkdir -p *) |
/scs-team:draft. Never auto-invoke based on project state or conversation context.You are helping the user create structured context for a specific concern area through conversation. Use this when they don't have existing documentation.
| # | Concern | Slug | What It Covers |
|---|---|---|---|
| 1 | Architecture | architecture | System structure, components, boundaries, tech stack |
| 2 | Security | security | Auth, data protection, threats, access control |
| 3 | Performance & Reliability | performance | Response times, availability, fault tolerance, scaling |
| 4 | Usability & Accessibility | usability | UX principles, accessibility, error handling patterns |
| 5 | Compliance & Governance | compliance | Regulatory requirements, audit, governance |
| 6 | Data & Provenance | data | Data models, data flow, retention, provenance |
| 7 | Testing & Validation | testing | Test strategy, coverage requirements, validation approach |
| 8 | Deployment & Operations | deployment | Infrastructure, CI/CD, environments, monitoring |
| 9 | Safety & Risk | safety | Risk assessments, safety checklists, mitigation strategies |
| 10 | Ethics & AI Accountability | ethics | AI usage policies, bias detection, audit trails |
| 11 | Business Context | business | Problem definition, stakeholders, objectives, success criteria |
Before asking questions:
.scs/ directory)Ask questions specific to the concern area. Be conversational, not interrogative.
Architecture Questions:
Security Questions:
Performance & Reliability Questions:
Usability & Accessibility Questions:
Compliance Questions:
Data Questions:
Testing Questions:
Deployment Questions:
Safety & Risk Questions:
Ethics & AI Accountability Questions:
Business Questions:
If answers are too vague, ask for specifics:
Based on the conversation, generate SCDs with version: "DRAFT".
Be honest about gaps:
Update concern bundle: After creating SCDs, update the relevant .scs/concerns/<concern>.yaml to include the new SCDs in its scds: array.
After creating/modifying SCDs, compile the .scs/ source to .claude/rules/ output:
.scs/scds/ and all concern bundles in .scs/concerns/.claude/rules/<concern-name>.md
d. Add header: <!-- scs-team:managed --> and footer with source SCD attribution<!-- scs-team:start --> / <!-- scs-team:end --> markers
b. If markers exist, replace content between them
c. If no markers, append the section at the end
d. If no CLAUDE.md, create one with the markers
e. Content: project overview, architecture summary, quick reference table, links to rules filesapplies_to.paths that reference sensitive data (PHI, PCI):
a. Add YAML frontmatter with paths: to the compiled .claude/rules/ file<!-- scs-team:managed --> header - leave other .claude/rules/ files untouchedAfter generating:
.claude/rules/User: /scs-team:draft architecture
You: "Let's document your architecture. I'll ask a few questions to understand the structure.
First - what's the high-level shape? Is this a monolith, microservices, serverless, or something hybrid?"
User: "It's a monolith right now but we're planning to break out some services"
You: "Got it - monolith with planned decomposition. What's the tech stack? Framework, language, database?"
User: "Python FastAPI, Postgres, Redis for caching"
You: "Nice stack. What are the main modules or domains within the monolith? The 3-5 big pieces?"
User: "Users, Orders, Inventory, and Notifications"
You: "And which of those are candidates for breaking out into services?"
User: "Notifications for sure, maybe Inventory"
You: "Last question - what external systems does this integrate with? Payment processor, shipping, etc.?"
User: "Stripe for payments, SendGrid for email, we're planning to add a shipping API"
You: "Got it. Let me generate the architecture context...
Generated SCDs:
system-context.yaml - Boundaries and external integrationscomponent-model.yaml - The 4 domains and their responsibilitiestech-stack.yaml - FastAPI, Postgres, Redis specificsUpdated Bundles:
concerns/architecture.yaml - Added 3 SCDsCompiled to Claude Code:
.claude/rules/architecture.md - Architecture rules and constraintsFlagged as TBD:
Suggested Next:
/scs-team:draft deployment - How this gets deployed/scs-team:draft data - Data models and flowRun /scs-team:status to see the full picture."