Claude SDLC Plugin Marketplace

A plugin marketplace for Claude Code spanning software development lifecycle automation, Gas Town bead-plan workflows, content generation (presentations, diagrams, documents), AI plugin testing, Copilot CLI development, and multi-agent team bootstrapping.

Quick Start

# Add the marketplace to Claude Code
/plugin marketplace add https://github.com/danielscholl/claude-sdlc

# Install any plugin
/plugin install sdlc
/plugin install gt-sdlc
/plugin install skilltest
/plugin install copilot
/plugin install cadre
/plugin install presentations
/plugin install excalidraw
/plugin install docx

Plugins

sdlc

Software development lifecycle automation with slash commands, specialized agents, and GitHub/GitLab webhook watchers.

Commands:

Command	Description
/sdlc:prime	Analyze and understand the codebase
/sdlc:feature [description]	Create a feature specification with deep codebase analysis
/sdlc:bug [description]	Create a bug fix specification
/sdlc:chore [description]	Create a maintenance task specification
/sdlc:implement [spec-file]	Implement a specification with structured task tracking
/sdlc:tdd [description]	Implement using strict Test-Driven Development
/sdlc:branch [spec-file]	Create a git branch from a specification
/sdlc:commit [spec-file]	Commit changes
/sdlc:pull_request [spec-file]	Create a pull request
/sdlc:pr_resolve [PR-number]	Resolve PR review comments in parallel
/sdlc:test_plan [module]	Analyze tests and create test documentation
/sdlc:locate [spec-file]	Locate specification files
/sdlc:install	Install dependencies and run setup scripts

Agents:

Agent	Purpose	Tools
architecture-strategist	Architectural review of code changes, design decisions, component boundaries	Read, Glob, Grep, Bash
performance-oracle	Algorithmic complexity, database queries, memory usage, caching, scalability	Read, Glob, Grep, Bash
pr-comment-resolver	Address PR review comments, implement fixes, post replies, resolve threads	Read, Edit, Write, Glob, Grep, Bash

All agents run on Sonnet for cost-efficient delegation.

---

skilltest

Four-layer test framework for validating Claude Code plugin skills remain well-maintained and purposeful.

Command: /skilltest:test-framework

Skill: test-framework — Validates plugin structure, tests trigger accuracy of skill descriptions, runs multi-turn session scenarios, and compares skill value (with vs without).

Testing Layers:

Layer	What it Tests	Speed
L1 Structure	Plugin spec compliance, naming, cross-references	~0.1s
L2 Triggers	Skill description accuracy (precision/recall)	~30s/query
L3 Sessions	Multi-turn routing, context handling, boundaries	2-3 min
L4 Value	Does the skill actually help? (with vs without)	5+ min

---

copilot

Create and maintain GitHub Copilot CLI plugins with authoritative spec references and templates.

Command: /copilot:copilot-plugin

Skill: copilot-plugin — Builds Copilot CLI plugin components: skills, agents, prompts, custom instructions, path-specific instructions, AGENTS.md routing, MCP servers, and hooks.

Includes reference specs for all Copilot CLI conventions, comparison matrices, a writing guide, and output templates.

---

cadre

Multi-agent team bootstrapper that analyzes your codebase and generates tailored agent teams via AI debate.

Command: /cadre:init

Skill: init — Analyzes project structure, runs an AI debate to determine optimal team composition, and generates .claude/agents/*.md files, config, and routing rules. Handles both existing projects and greenfield bootstrapping.

---

presentations

Branded presentation and LinkedIn carousel generation with a brand voice system and reusable slide layouts.

Skills:

Skill	Purpose
pptx-generator	Generate and edit PPTX slides, LinkedIn carousels, and reusable slide layouts
brand-voice-generator	Create tone-of-voice and brand-system files consumed by pptx-generator

---

excalidraw

Create .excalidraw diagram JSON files that make visual arguments — workflows, architectures, concepts.

Skill: excalidraw-diagram — Generates Excalidraw JSON with a visual-argument methodology and PNG render pipeline (requires Chromium via Playwright for rendering).

---

docx

Comprehensive DOCX document creation, editing, and analysis with tracked changes and comment support.

Skill: docx — Create new documents, modify content, work with tracked changes, add comments, extract text, and handle OOXML validation.

---

gt-sdlc

Gas Town SDLC pipeline — interactive brief → design → plan commands that turn ideas into Gas Town bead plans (bd create commands + dependency graph).

Commands:

Command	Description
/gt-sdlc:brief [verbiage]	Interactive product brief — conducts a dialog, writes docs/product-brief.md
/gt-sdlc:design	Interactive design doc — reads the brief, captures decisions, produces docs/design.md
/gt-sdlc:plan [design-path | verbiage]	Translate design.md or work-item verbiage into a bead plan (bd create heredocs + dep graph + convoy launch)

Designed to feed Gas Town — the multi-agent bead orchestration system. Each command produces reviewable markdown artifacts committed to docs/; the pipeline does not mask the Gas Town execution layer.

Testing

A root Makefile provides a unified test runner across all plugins.

# Run all fast tests (L1 lint + L2 trigger validation + pytest)
make test

# Target a specific plugin
make test P=copilot

# Individual layers
make lint                         # L1: Structure validation
make unit                         # L2: Trigger eval dry-run
make pytest                       # Unit tests for plugin scripts

# Slower tests
make integration                  # L3: Session tests (all plugins)
make integration P=sdlc           # L3: Session tests (one plugin)
make benchmark P=copilot S=copilot-plugin  # L4: Skill value comparison

# Inventory
make report                       # Test coverage across all plugins
make test-skill P=copilot S=copilot-plugin # All layers for one skill

Current coverage:

Plugin	L1 Targets	L2 Trigger Evals	L3 Scenarios	Unit Tests
sdlc	21	48 (3 agents)	1 (4 steps)	—
copilot	7	20	1 (4 steps)	43
skilltest	7	16	—	28

CLI Tool (Automation)

The sdlc CLI tool enables autonomous execution of SDLC commands via GitHub webhook watchers.

Installation

# Install from GitHub
uv tool install git+https://github.com/danielscholl/claude-sdlc

# Verify
sdlc --help
sdlc health

GitHub Watcher

Start the watcher to respond to GitHub issue comments:

sdlc watcher --port 8001

Then comment on any GitHub issue:

sdlc /feature add user profile page
sdlc /bug fix login error
sdlc please implement this          # AI classifies the task type
sdlc /feature add dark mode --plan-only

Workflows:

• Full automation: branch → plan → implement → commit → PR
• Plan-only (--plan-only): branch → plan → commit (stops for review)

See sdlc/README.md for CLI details and environment variables.

Development

Project Structure

claude-sdlc/
├── plugins/
│   ├── sdlc/              # SDLC automation plugin
│   │   ├── .claude-plugin/plugin.json
│   │   ├── agents/        # 3 specialized agents
│   │   ├── commands/      # 13 slash commands
│   │   └── tests/         # Trigger evals + session scenarios
│   ├── skilltest/         # Plugin testing framework
│   │   ├── skills/test-framework/scripts/  # L1-L4 test scripts
│   │   └── tests/         # Trigger evals + unit tests
│   ├── copilot/           # Copilot CLI plugin builder
│   │   ├── skills/copilot-plugin/
│   │   │   ├── reference/ # 9 authoritative spec docs
│   │   │   └── templates/ # 5 output templates
│   │   └── tests/         # Trigger evals + scenarios + unit tests
│   ├── cadre/             # Multi-agent team bootstrapper
│   │   └── skills/init/scripts/  # Analysis and generation scripts
│   ├── presentations/     # Branded PPTX and carousel generator
│   │   └── skills/        # pptx-generator, brand-voice-generator
│   ├── excalidraw/        # Excalidraw diagram creator
│   │   └── skills/excalidraw-diagram/
│   ├── docx/              # DOCX creation, editing, analysis
│   │   └── skills/docx/
│   └── gt-sdlc/           # Gas Town SDLC pipeline
│       └── commands/      # 3 slash commands (brief, design, plan)
├── sdlc/                  # Python CLI tool (webhook watcher)
├── tests/                 # CLI tool tests
├── Makefile               # Cross-plugin test runner
└── .claude-plugin/marketplace.json

Adding a New Plugin

1. Create plugins/{name}/.claude-plugin/plugin.json
2. Add skills, agents, or commands
3. Register in .claude-plugin/marketplace.json
4. Add trigger evals in plugins/{name}/tests/evals/triggers/
5. Run make test P={name} to validate

License

MIT