Javier Núñez

Project · open framework

GATE and the AI-assisted SDLC

Assistance that stops where judgement begins.

GATE, the foundation

GATE (Governed, Assisted, Traceable Engineering) defines the software life cycle for a team that works with AI assistants: the stages, what goes in and out of each one, who decides at each control point, and the line between what the AI does and what people do. The AI reads, correlates, drafts and checks against the source. The person accepts or rejects, chooses when something has to give, authorises every write into a shared system and answers for the outcome.

Its central piece is learning. Every decision taken in a review and every root cause of an incident becomes a rule in the team's catalog, with its origin written down, and that rule applies before the next change.

A change or an incident Review or postmortem Candidate rule, with its origin Custodian approves The team's catalog The team learns
The team's learning loop: every review and incident leaves a rule that applies before the next change

The SDLC, the house

Each company builds its own SDLC from GATE. That SDLC is a repository of its own, with its method, its rule catalog, the context of its products, its skills, its governance and the way each person gets in. It is written once in neutral formats, and a thin adapter wires it to the assistant the company chose, whether Claude Code, Gemini CLI, Kiro, Codex or another.

From one to the other

01 Name the orchestrator Who leads the adoption and decides 02 Choose the assistant Claude, Gemini, Kiro, Codex or another 03 Scaffold the skeleton One command writes the structure, the empty catalog and the guards 04 Interview About the company, its roles, its tools and what it wants first 05 Build in layers Each layer is shown before moving on 06 Verification Point by point, actually run 07 Each person gets in With their own access and their enabled profiles 08 Formulation with a written exit Once the criteria are met, each custodian approves their own scope
How to implement GATE in a company
python3 bin/bootstrap.py sdlc ../sdlc-mycompany --assistant claude-code

Tested in practice

GATE was put into practice in a real organisation, and what was learned there went back into the framework: the orchestrator's role, a formulation phase with a written exit, session permissions, and use cases that product validates and that become automated tests. It was also tested by building the SDLC of a simulated company from scratch and using it across the whole cycle, from defining a requirement to the postmortem of an incident.

Licence

The documentation is licensed CC BY 4.0 and the code MIT. You can use, adapt and redistribute it, commercially too, with attribution.

On the blog

  1. October 20, 2026

    How to implement GATE in your company

    What it takes to set up a company's SDLC with GATE, and where to start.

  2. October 13, 2026

    How GATE was built

    GATE started as a list of problems that kept repeating when working with AI.

  3. October 6, 2026

    Engineering with AI

    What a team gains when experience meets AI assistance and a cycle that learns.

Back to projects