Systems Modernization & AI Development

Implementing the spec: where discipline holds and where it slips

Right behaviour can ride on a wrong cited mechanism. Cross-artefact consistency is checked on apply, not on generation.

Posted on May 12, 2026

Post 1 and Post 2 covered the Constitution-generation phase and the audit moves that corrected it. Once the spec is stable, the same tool starts writing code, with the spec as the contract. [Read More]

Tags:

Auditing as you go: what corrections actually look like

Vague principles produce compliance theater. Sharp principles produce mechanical self-correction.

Posted on May 5, 2026

“Claude Code might be over-eager and add components that you did not ask for.” is the warning spec kit includes in its own README. [Read More]

Tags:

Spec Kit on a brownfield codebase: setup and first impressions

The interesting hallucinations show up in the Constitution, not the implementation.

Posted on May 3, 2026

In April and May I ran the same Spec Kit pipeline against two codebases with the same prompt. One was a small full-stack toy app I had refactored several times before the run, with intentional bugs left in place: sync sessions under async routes, dual ORM mapping, a hardcoded "not... [Read More]

Tags:

Quantifying a Legacy Codebase You Can't Rewrite

227 writes. Three semantic traps.

Posted on April 15, 2026

I inherited a 40-year-old codebase. [Read More]

Tags:

Conservative by Design: Transpiling Thousands of Legacy Scripts Without Guessing

A transpiler that flags what it can't do instead of guessing. Revolutionary, apparently.

Posted on March 25, 2026

I need to migrate several thousand scripts from a proprietary language into Python 3. The language is a Python 2.2 dialect with custom builtins bolted on over two decades. No documentation worth the name. No LLM has seen it in training. The scripts run a legacy ERP system, and the... [Read More]

Tags: