Typical AI assistant
- One general assistant for every role
- Context resets every conversation
- No process, protocols, or accountability
- Answers provided, delivery not included
etyb.ai assembles your complete virtual engineering company — from architecture to deployment. Senior expertise across every domain, every sprint, every stack.
The engineering team you're looking at runs on etyb-skills, our open-source, MIT-licensed skill — free today, and the foundation for what we're building next.
On Claude Code, install the plugin — it ships the skill plus the five model-tiered sub-agents and advisory hooks. That's the whole team. On any other agentskills.io-compliant agent, one npx command installs the same skill. Nothing to download or clone either way.
/plugin marketplace add e-t-y-b/etyb-skills/plugin install etyb@etyb-skillsSkill + 5 sub-agents (cheap models for cheap work, never above yours) + 5 advisory hooks.
npx skills add e-t-y-b/etyb-skills Codex, Cursor, Antigravity, Kiro, Cline + 14 more agents. Same knowledge, instruction-guided execution.
Either door installs the full knowledge tree — 14 specialists, 9 always-on protocols,
6 vertical architects, 13 vendor Stacks — as a single coordinated skill.
The plugin is what adds the execution layer on top: the sub-agents and hooks live in the
plugin manifest, so a skill-only install on Claude Code runs /etyb
without them. On Claude Code, take the plugin door.
Most AI tools give you an assistant. etyb.ai gives you a company — one orchestrator, 20 senior specialists, and 9 always-on protocols enforcing real engineering culture on every request. The team stays assigned through the plan, updates its own deliverables, and verifies its work before claiming done.
A feature, a system, a bug, a refactor. Natural language or technical spec. The orchestrator handles both.
The CTO skill. Determines which experts are needed, what protocols apply, and creates a living execution plan.
Domain experts take ownership of their areas. TDD first, verification before claims, review before merge.
Nothing ships without evidence. Tests pass. Code reviewed. Plan updated. You get working software.
etyb-skills is the open-source foundation we use internally. v4 collapsed
the architecture into one coordinated skill (/etyb)
backed by 20 specialists, 9 always-on protocols, 6 vertical architects, and 13 vendor Stacks —
all shipping in one repo, read from disk at runtime.
Portable by design. Runs on Claude Code (flagship, hook-enforced gates), OpenAI Codex, and Google Antigravity. Any agentskills.io-compliant platform, one install script. Inspect etyb. Read the protocols. Fork the references.
Stack Packs are knowledge overlays that load across your existing etyb team when work involves a specific platform. The team doesn't grow — it learns the platform. Same backend-architect writes the Lambda handler; same security-engineer enforces the IAM boundary. The overlay teaches them how to do it the way the platform expects, currency-stamped to a known date.
Three orthogonal axes: specialists + protocols + Stack Packs. v4 ships 13 vendor Stacks in-repo — no separate fetch, no docs site, no per-stack license.
Every role has defined expertise and process obligations. v4 collapsed the team into a
single skill (/etyb) with the references inside —
no peer skills competing at trigger time, no per-tier installs.
Click any node to inspect it. They stay assigned. They don't give up.
Start with etyb-skills. Open source, free, and ready to deploy. Or reach out to work directly with the etyb.ai team.