Systems

Agentic Ecosystem

An interconnected system of agents, skills, and tools — autonomously writing code, testing, deploying, and researching around the clock.

Featured Project

BigDawg AI Agent System

Three autonomous Claude-powered agents running 24/7 on dedicated hardware. They write code, deploy to production, run deep research, and coordinate through MCP servers and custom skills — no human in the loop.

View on GitHub

Agent Architecture

bigdawg.ts

BigDawg

Claude Haiku

Routing, tasks, triage

coder.ts

Coder

Claude Opus

Code, deploy, test

brain.ts

Brain

Claude Opus

Research, strategy, planning

Capabilities

Write, review, and refactor code via Claude Code

Deploy to Vercel and manage GitHub repos

CI/CD with Blacksmith.sh + Sprites.dev

Playwright testing and React prototyping

Multi-source deep research

MCP server development

Custom skill creation (SKILL.md)

Cross-agent coordination

Claude CodeAnthropic ClaudeOpenClawBlacksmith.shSprites.devVercelMCPTypeScriptGitHub Actions

Agentic Toolkit & Projects

Agentic Toolkit

Custom MCP servers, SKILL.md-based agent capabilities, and agentic workflow definitions. The composable infrastructure that powers autonomous development.

MCP ServersSKILL.mdClaude APITypeScriptTool Use

AI-Native Test Suite

End-to-end tests written in plain English, powered by Shortest and Claude. Agents validate their own output — no manual test authoring required.

ShortestPlaywrightAnthropic ClaudeNatural LanguageE2E

romanchursin.dev

This portfolio — built entirely through Claude Code with zero manual coding. CI/CD via GitHub Actions, Blacksmith.sh runners, and Sprites.dev optimization.

Claude CodeNext.js 16React 19Blacksmith.shVercel

OpenClaw Setup

Multi-agent system configuration with agent identity definitions, skill registries following the SKILL.md standard, and cost optimization via prompt caching.

OpenClawClaude APISKILL.mdMCPTelegram