Skip to content
AvailablePT

Professional and client work

AI development framework

Internal Liquid AI framework on top of Claude Code, with the test before the code and gates for lint, types, tests and build. Three models have separate roles: one implements, another reviews, and the orchestrator validates and commits.

The problem

A team that uses coding agents without a shared standard ships fast but without control: each person works a different way, and nothing stops a delivery without tests or an action without consent.

Technical decisions

Mandatory TDD and gates for lint, types, tests and build before each delivery. Specialized agents, reusable automations, blocking hooks and layered persistent memory. Inviolable rules for security, human consent and a ban on autonomous commits by agents.

Outcome

It sustained the team's software delivery at Liquid AI, with MCP integrations and a custom memory server shared between agents. The code is not public.

Work under confidentiality: the code and the URL are not public.

  • Claude Code
  • MCP
  • TDD

Ready to share.

Let’s talk about your project?

The person answering is the person building it. Tell me what you need and I will explain the simplest path.

I prefer to talk on WhatsApp