Uses
My stack, my tools and, above all, how I work with AI agents day to day.
How I work with AI agents
I cover my development lifecycle end to end with agents: moving from a traditional SDLC to an AIDLC (AI-Driven Development Lifecycle). An orchestrator agent hands work to specialists by phase; I set the goal and review the results.
Orchestratortech-leadMy orchestrator agent: takes the goal, breaks it down and hands the work to the specialist agents.
Define
- spec-writerSpecs and acceptance criteria.
- business-docBusiness documentation.
Design
- architectArchitecture and technical decisions.
- plannerStep-by-step implementation plan.
Build
- tdd-guideTests first.
- backend-engineerImplementation.
Verify
- code-reviewerCode review.
- security-reviewerSecurity review.
- qa-engineerQuality and end-to-end testing.
Document
- doc-updaterKeeps the docs up to date.
Agents, orchestration and skills
- Claude Code My main agent to write, review and verify code.
- Codex Parallel tasks and a second opinion.
- OpenCode Another coding agent in the terminal.
- Orca Orchestrating agents in parallel.
- Herdr Agent orchestration, mainly.
- AI Hero skills Reusable skills for the agents.
- caveman Short, direct answers: fewer tokens, same substance.
- ponytail The simplest solution that works. YAGNI first.
- impeccable Interface design skill.
- archify Interactive architecture diagrams.
- RTK Cuts the tokens command output consumes.
- ccusage To see agent usage and cost.
What this site is built with
Built with agents working in parallel, each with its own contract and files, and verified with tests, screenshots and Lighthouse.
- Astro Static site, MDX content.
- Three.js The khipu, the arcade and the 3D world.
- Bun Install, scripts and tests.
- TypeScript All the code.
- Biome Format and lint.
- Playwright End-to-end tests and screenshots.
- Lighthouse CI Core Web Vitals on every PR and in production.
- GitHub Actions + Pages CI and deploys.
- Umami Cookieless analytics.
- Web AudioMusic and sounds generated in the browser.
Editor, terminal and gear
Professional stack
What I use at work, by area.
- Languages
- TypeScriptJavaScriptPythonGoRustKotlinJava.NET
- Cloud
- AWSGCPAzureKubernetesKarpenterTerraformPulumiHelm
- Platform
- KongKafkaArgoCDGitHub ActionsGrafanaPrometheusLokiTempo
- Architecture
- MicroservicesServerlessEDAMicro frontendsMACHC4
- Data
- PostgresDynamoDBRedisMongoDBRedshiftOpenSearch