SDLC orchestration plugin for Claude Code. Drives projects through a structured 10-phase software development lifecycle with company-configurable profiles, compliance gates, and quality enforcement.
No existing tool combines specification-driven development + quality enforcement + compliance gates + company configurability into one executable workflow. This plugin makes the AI-SDLC methodology executable inside Claude Code.
- 10 SDLC phases — Discovery through Monitoring, each with entry/exit gates
- Project type routing — Phases 6–9 adapt automatically based on
project_type(service, app, library, skill, cli) - Company profiles — YAML configs for stack, quality thresholds, compliance, conventions
- 6-gate validation — Integrity, completeness, metrics, compliance, consistency, quality checks at every transition
- Frozen layers — Token-efficient phase summaries enabling cross-phase context continuity
- 3-tier context architecture — Foundation (always), frozen layers (per-phase), references (on-demand)
- Narrative enhancement — Optional stakeholder-friendly
.narrative.mdcompanions for technical artifacts - Conversational coaching — Adaptive dialogue mode for guided phase completion
- NFR measurement basis enforcement — Every numeric threshold must declare its measurement source; aspirational thresholds require an agreed Phase 6 validation plan
- Test traceability — Phase 6 requires scenario-to-requirement mapping and redundancy audit before test execution
- Compliance enforcement — SOC 2, HIPAA, GDPR, PCI-DSS gate definitions
- Skill orchestration — Maps phases to existing skills (
/deep-plan,/deep-implement,/tdd,/code-review,/security-review,/e2e) - State machine — Tracks progress in
.sdlc/state.yamlwith full audit trail - Artifact checksums + dirty tracking — SHA-256 hashes detect changed artifacts; gate checks skip unchanged files for faster validation
- Smart repair — Gate failures on structural issues (missing templates, placeholders) are auto-fixed before escalating to human
- Dependency graph enforcement — Section dependency DAG validated in Phase 4; circular dependencies detected; implementation order enforced
- Multi-perspective review —
/sdlc-reviewwith adversarial, edge-case, and council modes for thorough artifact review - Brainstorming techniques — SCAMPER, reverse brainstorming, constraint removal, and 3 more structured ideation methods with anti-bias protocol
- Brownfield detection — Auto-detects existing codebases and generates workspace analysis before Discovery begins
- Structured error specs — P0/P1 requirements must define Accepts/Returns/Errors explicitly
- Cross-artifact validation — File-level reference checking between artifacts (SHOULD severity)
- Question-to-file audit trail — Open questions written to persistent markdown files for team collaboration
- Test expansion protocol — Healer (diagnose failures), Expander (add edge cases), bug-to-test pipeline (RED first)
- Parallel review resolution — Batch review findings by file, resolve in parallel, re-verify after fix
- Advanced elicitation — Constraint analysis, inverse thinking, pre-mortem, 5 whys, and more for coaching mode
- Phase-aware context — Hooks inject current phase reminders and conventions
# Clone the repo
git clone https://github.com/MCKRUZ/claude-code-sdlc.git
# Symlink to your Claude Code skills directory
# Windows
mklink /D "%USERPROFILE%\.claude\skills\claude-code-sdlc" "path\to\claude-code-sdlc"
# macOS/Linux
ln -s /path/to/claude-code-sdlc ~/.claude/skills/claude-code-sdlcThe plugin uses Python scripts via uv:
# Install uv (if not already installed)
pip install uv
# or
brew install uv
# Install script dependencies
cd claude-code-sdlc/scripts
uv sync1. /sdlc-setup → Select profile, initialize .sdlc/ in your project
2. /sdlc → See Phase 0: Discovery guidance
3. Create artifacts → Write problem-statement.md in .sdlc/artifacts/00-discovery/
4. /sdlc-gate → Check if exit criteria are met
5. /sdlc-next → Advance to Phase 1: Requirements
6. /sdlc-status → View progress dashboard
For in-depth technical documentation, see the guides in docs/:
| Guide | Description |
|---|---|
| Architecture | Plugin anatomy, component relationships, data flow diagrams, two-directory model, progressive disclosure strategy |
| Phase Lifecycle | All 10 phases in depth — workflows, artifacts, HITL gates, skills, agents, handoff protocol, project type adaptations |
| Gate System | 6-gate validation — integrity, completeness, metrics, compliance, consistency, quality — severity levels, override protocol |
| Profiles | Schema reference (every field), built-in profiles, custom profile creation, compliance framework integration, evaluation criteria |
| Commands | All 11 slash commands — internal flow, state changes, Python scripts called, error scenarios, examples |
| Agents | 7 custom agents + built-in subagent orchestration, phase-to-agent mapping, parallel execution rules, mandatory spawns |
| State Machine | state.yaml format, transition rules, history tracking, session-handoff.json, sections-progress.json |
| Templates & Artifacts | Template directory structure, per-phase artifact details, handoff document protocol, artifact lifecycle |
| Scripts | All 12 Python scripts — CLI args, inputs/outputs, exit codes, gate implementation details, uv runtime |
| Integrations | How /deep-plan, /deep-implement, /tdd, /code-review map into SDLC phases, artifact transformation pipeline |
| Hooks | Session-start and phase-inject hooks — what they read, what they inject, session continuity, convention reminders |
| Command | Purpose |
|---|---|
/sdlc-setup |
Interactive setup wizard — select profile, initialize project |
/sdlc |
Show current phase guidance, next action, required artifacts |
/sdlc-status |
Progress dashboard with phase table and completion % |
/sdlc-gate |
Run 6-gate exit criteria check (does not advance) |
/sdlc-next |
Advance to next phase if all MUST gates pass |
/sdlc-enhance |
Generate narrative companions for stakeholder review (optional) |
/sdlc-coach |
Interactive coaching mode — adaptive dialogue for current phase |
/sdlc-phase-report |
Generate phase HTML report with artifact inventory |
/sdlc-review |
Multi-perspective artifact review (council, adversarial, or edge-case modes) |
/sdlc-audit |
Analyze gate effectiveness across completed phases |
Full enterprise stack with compliance:
- Stack: C#/.NET 8, Angular 17, SQL Server, Azure
- Quality: 80% coverage minimum, 100% critical paths, TDD required
- Compliance: SOC 2 gates at every phase transition
- Conventions: Conventional commits, immutable patterns, no console.log
Minimal profile for quick start:
- Stack: Configurable (defaults to TypeScript/Node)
- Quality: 60% coverage minimum, no TDD requirement
- Compliance: None
- Conventions: Conventional commits
Copy profiles/starter/profile.yaml and modify for your stack. Validate with:
uv run scripts/validate_profile.py profiles/my-profile/profile.yamlSee profiles/_schema.yaml for the full schema.
Phase 0 captures project_type, which is stored in state.yaml and controls how Phases 6–9 execute:
| Type | Description | Phase 6 | Phase 8 | Phase 9 |
|---|---|---|---|---|
service |
Backend API / server process | Unit + integration + E2E + API contract tests | Staging server deploy + DB migrations | Full infrastructure monitoring + alerting |
app |
User-facing application with UI | Same as service | Same as service | Same as service |
library |
Shared code package / SDK | Unit tests for public API surface | Package registry publish | Download counts + issue triage |
skill |
Claude Code skill / AI plugin | Scenario-based testing against requirement list | File distribution + install verification | Qualitative feedback channels + issue triage |
cli |
Command-line tool | Unit + integration + invocation tests | Binary distribution | Same as library |
| # | Phase | Primary Skills | Key Artifacts |
|---|---|---|---|
| 0 | Discovery | /plan |
problem-statement.md, constitution.md (incl. project_type) |
| 1 | Requirements | /deep-project |
requirements.md, non-functional-requirements.md (with Measurement Basis) |
| 2 | Design | /deep-plan |
design-doc.md, ADRs |
| 3 | Planning | /deep-plan |
section plans |
| 4 | Implementation | /deep-implement, /tdd |
source code, tests |
| 5 | Quality | /code-review, /security-review |
review reports |
| 6 | Testing | /e2e, /test-coverage |
test plan (with traceability map), coverage report |
| 7 | Documentation | /update-docs |
README, RUNBOOK |
| 8 | Deployment | CI/CD | release notes (approach varies by project_type) |
| 9 | Monitoring | Manual | monitoring config (approach varies by project_type) |
Every phase transition runs through six gates:
- Integrity — Required artifacts exist and are well-formed
- Completeness — Artifacts contain all required sections, no placeholders
- Metrics — Quantitative thresholds met (coverage, file size)
- Compliance — Correct labeling (priorities, ADR status, compliance mapping)
- Consistency — Locked metrics checked against frozen layers from prior phases
- Quality — Holistic assessment (clarity, accuracy, consistency)
Gates have severity levels:
- MUST — Blocks transition if failed
- SHOULD — Generates warning
- MAY — Informational
claude-code-sdlc/
├── plugin.json # Plugin manifest
├── SKILL.md # Main skill entry point
├── commands/ # 11 slash commands (/sdlc, /sdlc-setup, /sdlc-status, /sdlc-next, /sdlc-gate, /sdlc-enhance, /sdlc-coach, /sdlc-review, /sdlc-phase-report, /sdlc-audit)
├── agents/ # 7 agents (orchestrator, requirements-analyst, compliance-checker, section-evaluator, narrative-enhancer, gate-repair, multi-reviewer)
├── profiles/ # Company/stack YAML profiles
├── phases/ # Phase definitions (00–09)
├── references/ # Progressive disclosure docs
├── templates/ # Artifact templates
├── scripts/ # Python automation (uv)
└── hooks/ # PowerShell hooks for context injection
This plugin follows ADR-001: claude-code-sdlc is a Claude Code plugin that executes AI-SDLC methodology.
- AI-SDLC = the methodology (phases, validation, templates)
- claude-code-sdlc = the execution engine (commands, hooks, profiles, scripts)
- The plugin references AI-SDLC patterns but adapts them for Claude Code's command/hook/agent system
- Fork the repo
- Create a feature branch:
feat/my-feature - Follow the conventional commit format:
type: description - Submit a PR with a clear description
MIT