AI enablement
The Squad Starter Pack, equipping a team to work with AI
A set of files and skills placed in the team's repository gives the AI the context it needs, durably, shared across sessions and across people. Four foundations first, contextual extensions next, and a self-check that says where the team stands and what to start with. The goal is not compliance: it is creating the conditions for the AI to be genuinely useful, durably and collectively.
The short version of the story: this pack was not designed as a framework, it was extracted from a real project. During the rebuild of a product portal in design-to-code, every guardrail was put in place in response to a real problem, from the initial false start (too much context injected at once, the AI improvising) to the three-layer system that holds without manual vigilance. The full story of that project, numbers included, is in the note Directing AI, not coding in its place.
What changed next was the shift from “my practice” to “our equipment”. The shared version of the pack was co-validated with product and tech, precisely so it would not be a designer’s standard imposed on the other functions. Each dimension has a natural owner: the product file belongs to the PM, the design document to the designer, the conventions to tech. That co-ownership is what holds the whole thing together.
Since the summer of 2026, the pack has been living its real test: its creator has left, the framework and the instances remain. That is exactly what it was designed for.
The complete framework is on this page, deliberately: an enablement method that hides enables nobody. What stays private is the internal instantiation, the real files and the teams’ scores. If you want to see what a pack looks like in place in a real repository, ask: it can be shown live, gladly.