From rptc
Planning methodology for the RPTC architect-agent. Covers incremental sub-agent delegation, implementation constraints generation (5 categories), 6-phase planning methodology (context analysis, test strategy, implementation steps, dependencies/risk, self-critique, output), context management strategy, quality standards, best practices, and anti-patterns. Produces TDD-ready implementation plans.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rptc:architect-methodologyThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Planning methodology for the RPTC architect-agent. Produces comprehensive, TDD-ready implementation plans.
Planning methodology for the RPTC architect-agent. Produces comprehensive, TDD-ready implementation plans.
Create a comprehensive, TDD-ready implementation plan that guides developers through test-driven development with clarity, completeness, and actionability.
Philosophy: Tests MUST be designed BEFORE implementation steps. The plan is the contract between idea and execution.
Directory Format Only (v2.0.0+): Uses incremental sub-agent delegation for all features.
The plan command delegates to specialized sub-agents:
Output: Directory structure with multiple files
~/.claude/plans/[feature-slug].md - Native plan file (Claude's plan mode)Advantages:
For EVERY plan, analyze the project and generate all 5 categories:
.context/ files and codebase for file size patterns.context/project-overview.md for tech stackGenerate constraints in overview.md. Keep concise (100-200 lines max).
Load and analyze all relevant context:
Standard Operating Procedures - Consult ${CLAUDE_PLUGIN_ROOT}/sop/[name].md:
architecture-patterns.md, testing-guide.md, languages-and-style.md, security-and-performance.mdProject Overrides - Check .context/ directory for project-specific requirements
Research Integration - Extract recommendations from provided research documents
Codebase Patterns - Search for similar implementations, testing patterns, naming conventions
Before writing ANY implementation steps, design the complete test strategy.
2.0 Specification Collection across 6 areas:
2.1 Test Types Required: Unit, Integration, E2E
2.2 Test Scenarios: Happy path, edge cases, error conditions (Given-When-Then format)
2.3 Test Design Patterns: Given-When-Then (BDD), Arrange-Act-Assert (AAA), Boundary Value Analysis, Equivalence Partitioning
2.4 Coverage Goals: Overall 80%+, critical paths 100%, error handling all paths
For each step, provide:
Step Ordering: Dependencies first → Highest risk → Core functionality → Testability
Checkbox format required: All tasks use - [ ] for TDD phase tracking.
Before finalizing, perform comprehensive self-critique:
Completeness Check: All acceptance criteria, test coverage (happy/edge/error), dependencies, risks
Simplicity Validation (KISS/YAGNI Gate):
| Metric | Threshold | Action if Exceeded |
|---|---|---|
| Implementation steps | ≤10 | Merge related steps |
| New files | ≤5 | Consolidate files |
| Layers of indirection | ≤3 | Remove intermediate layers |
| Abstractions | ≥3 use cases each | Inline single-use abstractions |
Quality Scoring (1-5 scale, target ≥ 4.0):
Generate comprehensive plan following the output template. See references/output-template.md for exact structure.
Token Budget Target: <15K tokens for plan output
Your plan is successful when:
/rptc:feat or /rptc:fix command) — always, regardless of plan complexity- [ ] for tracking- [ ]For detailed templates and examples:
${CLAUDE_PLUGIN_ROOT}/skills/architect-methodology/references/output-template.md - Complete plan output template with all sections${CLAUDE_PLUGIN_ROOT}/templates/plan-overview.md - Plan overview format${CLAUDE_PLUGIN_ROOT}/templates/plan-step.md - Step detail formatnpx claudepluginhub kuklaph/rptc-workflow --plugin rptcGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.