Open-Source Framework Lattice Tackles AI Coding Assistant Flaws, Enforcing Engineering Discipline

A new open-source framework called Lattice aims to fix the critical failures of AI coding assistants, which often bypass engineering standards, forget constraints, and produce unreviewed code. Created by Rahul Garg, the framework introduces composable skills—atoms, molecules, and refiners—that embed battle-tested practices like Clean Architecture, Domain-Driven Design, and secure coding. A living context folder (.lattice/) accumulates project standards, decisions, and review insights, allowing the system to learn and apply project-specific rules over time.

Announced today, Lattice can be installed as a Claude Code plugin or used independently with any AI tool. Garg emphasized the urgency: “AI coding assistants jump straight to code, silently make design decisions, forget constraints mid-conversation, and produce output nobody reviewed against real engineering standards. Lattice fixes this with composable skills that embed battle-tested engineering disciplines.” The framework is the culmination of a series of posts Garg published over the past two months on reducing friction in AI-assisted programming.

Background: The Problem with Current AI Assistants

The rise of AI coding tools has exposed a dangerous gap: they generate code quickly but often neglect architectural integrity and security. Garg’s previous writings highlighted how assistants make silent design choices, lose track of constraints, and output code that bypasses peer review. Lattice directly addresses these issues by enforcing structured development patterns.

Open-Source Framework Lattice Tackles AI Coding Assistant Flaws, Enforcing Engineering Discipline — Source: martinfowler.com

In a related development, an article by Wei Zhang and Jessie Jie Xia on Structured-Prompt-Driven Development (SPDD) has generated massive traffic. The authors have now added a Q&A section answering a dozen common questions, clarifying how SPDD works in practice. This addition aims to help developers adopt more disciplined prompt engineering.

What This Means: A Shift Toward Developer-Controlled AI

Lattice represents a significant shift—from AI-as-black-box to AI-as-extensible-framework. Developers can now mold their AI assistant to follow specific engineering disciplines, effectively enabling internal reprogrammability. This concept, central to Smalltalk and Lisp communities but lost in modern IDEs, returns with AI making software change superfast.

Jessica Kerr, a prominent developer, posted about building a tool to work with conversation logs, observing a double feedback loop. “There are (at least) two feedback loops running here. One is the development loop, with Claude doing what I ask and then me checking whether that is indeed what I want… Then there’s a meta-level feedback loop, the ‘is this working?’ check when I feel resistance.” Her tool helps developers change not only their code but also the environment they use to build it—a meta-loop that pays off immediately.

Kerr added: “As developers using software to build software, we have potential to mold our own work environment. With AI making software change superfast, changing our program to make debugging easier pays off immediately. Also, this is fun!” This sentiment echoes Garg’s vision for Lattice: a framework that grows smarter with use, applying your rules, informed by your history.

Together, these developments signal a new era where AI programming tools become customizable, disciplined, and transparent—restoring the lost joy of shaping one’s development environment to exactly fit the problem.

Tags:

Open-Source Framework Lattice Tackles AI Coding Assistant Flaws, Enforcing Engineering Discipline

Background: The Problem with Current AI Assistants

What This Means: A Shift Toward Developer-Controlled AI

Related Articles

Recommended

Discover More