davekim917-bootstrap

bootstrap

A portable agent environment for Claude Code. Four independent plugins — install what you need.

Built for my own use, shared in case it's useful to others.

Plugins

Plugin	What it provides
bootstrap-workflow	Team workflow skills (brief → design → review → plan → build → qa → ship), safety/quality hooks, specialized agents
bootstrap-domain	Domain expertise skills for software engineering, data, analytics, AI, and finance
bootstrap-commands	Codebase analysis commands that generate CLAUDE.md, project skills, and AI dev setup
bootstrap-tools	Tool integration skills for CLI tools (Cortex Code, and more)

Install

Add the marketplace, then install the plugins you want:

/plugin marketplace add davekim917/bootstrap

Install only what you need:

/plugin install bootstrap-workflow@davekim917-bootstrap     # team workflow + hooks + agents
/plugin install bootstrap-domain@davekim917-bootstrap       # domain skills only
/plugin install bootstrap-commands@davekim917-bootstrap     # bootstrap commands only
/plugin install bootstrap-tools@davekim917-bootstrap        # tool integration skills

Prerequisites

• Claude Code CLI
• Bun runtime (for TypeScript hooks — workflow plugin only)
• jq (for shell hooks — brew install jq on macOS)

bootstrap-workflow

13 team-* skills, 6 agents, safety/quality hooks.

Skills

Skill	Purpose
/team-brief	Extract structured requirements from fuzzy ideas
/team-design	First-principles design with constraint analysis
/team-review	Adversarial multi-model design review
/team-plan	Atomic task decomposition with exact file paths and test cases
/team-build	Parallel build with coordinated builder agents
/team-qa	5-check validation pipeline (denoise, style, docs, security, perf)
/team-ship	Branch lifecycle and merge/PR options
/team-tdd	Test-driven development enforcement
/team-debug	Root-cause-first debugging methodology
/team-drift	Mechanized drift detection between any two documents
/team-retro	Post-ship learning capture
/team-verification-before-completion	Evidence-based completion claims
/team-receiving-review-feedback	Protocol for processing review findings

Agents

Specialized subagents invoked by workflow skills:

• architecture-advisor — architectural review and design feedback
• code-review-specialist — code quality and convention review
• cpo-advisor — product strategy and prioritization
• cto-advisor — strategic technical decisions
• performance-analyzer — performance issue detection
• security-reviewer — security vulnerability review

Hooks

Safety (PreToolUse):

• Block destructive bash commands (rm on protected paths, eval, shred, unlink)
• Block edits to protected files (.env, .git/, lock files, terraform)
• Workflow gate enforcement (blocks build without passing drift check)

Quality (PostToolUse):

• TypeScript type checking after file edits
• Prettier formatting after file edits
• Edited file tracking for session-end review

Lifecycle (SessionStart):

• Session timestamp injection
• Cloud MCP connector management

Usage

For a non-trivial feature:

/team-brief → /team-design → /team-review → /team-plan → /team-build → /team-qa → /team-ship

For smaller work with clear requirements, skip to /team-design → /team-plan → /team-build.

For trivial fixes, skip the workflow entirely.

bootstrap-domain

8 domain expertise skills — loaded contextually based on your project:

• software-engineering — TypeScript, React, Node.js, APIs, testing
• analytics-engineering — dbt, SQL modeling, data transformation
• analytics — dashboards, metrics, BI tools
• data-engineering — pipelines, orchestration, data quality
• data-science — notebooks, ML, feature engineering
• ai-integration — LLM APIs, prompt engineering, RAG, agents
• financial-analytics — GL modeling, reconciliation, regulatory reporting
• jony-ive — premium UI/UX design audit and refinement

bootstrap-commands

7 commands for bootstrapping AI-assisted development in any codebase:

Command	Purpose
/bootstrap	Orchestrator — guides you through the full sequence
/bootstrap-discovery	Stage 1: Analyze codebase patterns and architecture
/bootstrap-config	Stage 2: Generate CLAUDE.md and AGENTS.md
/bootstrap-skills	Stage 3: Generate project-specific skills
/bootstrap-domain	Stage 4: Generate domain-specific skills
/bootstrap-audit	Stage 5: Audit and reconcile all artifacts
/bootstrap-complete	Stage 6: Final cleanup and validation

Also includes the skill-developer utility skill for creating new skills.

Structure