05 / system / framework

Clause: an intent compiler for AI-assisted software.

Clause is an experimental framework for building software with AI doing much of the writing, without losing control of the design. It lowers human intent through structured, checkable artifacts until code can be written, reviewed, and repaired against what it was meant to satisfy.

KindCreative work / framework
Developed byYoav Fekete
DomainAI-assisted software development
StatusExperimental
Public framingIntent compiler

05.1 / what it is

A discipline for keeping intent connected to code.

Clause is a way to build software with AI doing most of the writing, without letting the design dissolve into a pile of plausible edits.

You can think of it as an intent compiler. Not a compiler that magically turns an idea into finished code, but a system that lowers human intent through a chain of more concrete forms: from what the system should do, to how it should be shaped, to the commitments each code unit must satisfy, and finally to code.

At each level, the same loop repeats: one worker writes from the level above, and another checks the result before work moves downward. Write, then check. Write, then check. All the way to code.

The important part is that the code does not float free. Each relevant piece of code carries a small trace back to the promise it is meant to satisfy. The documents say what should exist. The code shows what has actually been built against those documents.

System identity
Name
Clause
Kind
Artifact-discipline framework
Public metaphor
Intent compiler
Substrate
Repository-native artifacts
Primary concern
Design intent
Validation stance
Deterministic where possible
Status
Experimental / in development
Relationships

05.2 / why now

AI made code generation cheap. Preserving meaning became the hard part.
Thesis

AI can write code faster than humans can manually preserve the meaning of those changes. The hard question is no longer only “can the code be generated?” It is whether the code still realizes the architecture, assumptions, invariants, and decisions that gave the system meaning.

Design intent must survive the edit loop.

The danger is drift. A change can look reasonable by itself and still quietly break an assumption made somewhere else. One session adds a small mismatch. The next adds another. After enough sessions, the system may still run, and the code may still look plausible, but the design has started to dissolve.

The obvious answer is to review everything more carefully. But that does not scale. Not having to read every line yourself is one of the reasons to use AI in the first place.

05.3 / the key move

Make context computable instead of manually chosen.

A second AI reviewer can help, but only if it sees the right context. Too little context, and it misses the real issue. Too much context, and it drowns. The valuable human work is not copying text between agents. It is deciding what context matters.

Clause tries to remove the human from that handoff. Not by guessing better. Not by search. Not by asking a model what seems relevant. The key move is to make context something the system can compute.

To work on one piece, an AI needs to know what that piece is supposed to do and what the pieces around it promise. It usually does not need their full implementation. It needs their public commitments: what they guarantee, what they require, and what must not be broken.

Those neighboring pieces are not chosen by intuition. They follow from the connections in the project. So the connections must be written down in a form a program can follow. Once that is true, the right context can be gathered the same way every time.

What Clause computes
Starting point
The unit being written or reviewed
Local contract
What this unit must satisfy
Neighbor contracts
What nearby units promise
Architecture context
The design intent that constrains the change
Excluded by default
Irrelevant implementation bodies

05.4 / why boundaries matter

Automatic context needs pieces with clear starts and ends.

For automatic context to work, the system must be able to lift out the right pieces cleanly. That means every important piece of meaning needs a clear start and a clear end.

A marker that only says “look here” is useful for navigation, but it is not enough for context. It points to a place; it does not say what content belongs to that place.

Clause therefore treats important definitions, commitments, rules, and design entities as bounded repository artifacts. If something may need to be handed to an AI, the system must know exactly where it begins and exactly where it ends.

This is the small mechanical rule that makes the larger loop possible.

05.5 / artifact chain

Intent travels through artifacts, not chat memory.
01

Intent

What the system is meant to do and why it matters.

02

Design

How the system is shaped into entities, rules, and relationships.

03

Contract

The commitments a specific code unit must satisfy.

04

Code

The implementation, carrying a trace to what it realizes.

05

Evidence

Tests, static checks, validation records, or review results.

05.6 / the loop

Writing, mechanical checking, and independent review each do different jobs.
Layer What it does Why it matters
Writer Produces the next artifact or code change from the level above. The writer works inside a bounded context instead of inventing the architecture from chat memory.
Mechanical checker Catches failures with definite answers: broken links, missing pieces, malformed markers, unresolved references, structural gaps. Anything deterministic should be checked by a tool, not remembered by a person.
Independent reviewer Judges what a machine cannot decide: whether the change really satisfies the intent. The reviewer sees the contract and the change, not the writer's self-justification.
Person Directs the system, resolves judgment calls, and shapes the architecture. Clause does not remove human design judgment. It moves the person upward.

05.7 / neighboring categories

Clause sits around AI-assisted coding rather than inside a coding assistant.
Neighboring category What it does How Clause differs
AI coding agents Generate or modify code. Clause gives agents bounded contracts and surrounding intent, so generated code remains answerable to the design.
AI code review tools Review diffs and enforce rules. Clause makes the relevant intent and contracts first-class repository artifacts before the diff reaches review.
RAG / semantic retrieval Find likely relevant context through search or similarity. Clause aims for computed context: graph traversal over explicit relationships, not guessed relevance.
Spec-driven tools Create specs or plans before implementation. Clause emphasizes traceability through implementation and later edits, not only planning before code exists.
Formal methods Prove specific properties. Clause tracks where commitments come from, how they are realized, and which tests, checks, or reviews currently support them.

05.8 / current status

A framework/specification/tooling direction in active development.

Clause is experimental. Current work centers on the artifact discipline itself: the reference system, manifest model, doctrine documents, deterministic validator, context assembly, and orchestration model that let human judgment, LLM assistance, and mechanical checks cooperate without confusing their roles.

The framework is being tested against architecture-heavy robotics software, where the cost of design drift is high and many facts can be represented as requirements, invariants, geometric constraints, implementation patterns, tests, and review obligations. That setting is a proving ground for the discipline, not a claim that Clause is a released product.

Status boundaries
Stage
Experimental
Public package
Not claimed
Public repo
Not claimed
Production readiness
Not claimed
Schema type
CreativeWork

05.9 / relationship to Yoav's work

A practical response to complex software and AI-assisted editing.

Clause grew out of Yoav Fekete's work with complex robotics software and AI-assisted development. The practical problem was not that LLM agents were useless; it was the opposite. They became useful enough to accelerate implementation, and powerful enough to blur architecture, erase design commitments, or turn long-term intent into disposable chat context.

In the broader narrative of this site, Clause is one way of making a problem take the right form. It represents software intent as repository artifacts that humans, agents, and deterministic tools can inspect, address, and check. That keeps Clause connected to Yoav's wider work in structured systems without collapsing Clause into the person, the robotics domain, or any single coding tool.

05.10 / related pages

Public routes that place Clause in the site graph.