Writing / Field notes

Working ideas, made inspectable.

Notes from building products, leading technical systems, recovering difficult programs, and making AI-assisted engineering more governable.

14 published notes

Search articles · Subscribe with RSS

01

Resolve conflicting requirements before implementation

Use an export-permissions example to separate assumptions, existing behavior and the policy decision someone must make.

02

Know which tests cover which requirements without a spreadsheet

Check one export-permission requirement against a specific test assertion, record the evidence, and revisit the link when either changes.

03

When every fix makes something else worse, look for the contradiction

Use a cache-size example to turn opposing requirements into a design candidate, then test it against both response time and memory constraints.

04

Check how an agent handoff could fail

Trace a changed dependency response to an unreadable delivery report, then distinguish planned controls from controls you have tested.

05

Plan the work your AI agents do like a project, not a to-do list

Use a documentation handoff to make dependencies explicit and keep schedule estimates distinct from observed delivery.

06

Keep the reasoning behind an agent's result

Keep a coding agent’s claim, prediction, observations and conclusion together so another reviewer can reconstruct the decision.

07

Your coding agent's "done" is a hypothesis. Treat it like one.

Use a webhook-retry example to separate the agent’s completion claim, the expected behavior and the evidence needed to accept it.

08

Build a small lab for your browser-testing tool

Build a controlled fixture and answer key, retain incomplete runs, and test the scorer before using it to compare browser-testing tools.

09

Before building a roadmap item, test the premise behind it

Use one proposed parallel runner to distinguish a plausible feature from a problem your records can actually establish.

010

Plant the bugs yourself: how to find out if your testing tool cries wolf

Use planted defects and working cases to distinguish useful findings from false alarms, with a small Jevitate evaluation and its limits.

011

Keep end-to-end tests useful as your app changes

When a browser test breaks after a screen changes, preserve the user outcome before repairing the steps. A worked order-flow example shows what to check.

012

Write a coding-agent handoff you can review

Use a bounded slug-helper example to define scope, acceptance checks and the evidence needed before applying a change.

013

Catch a failed save even when the page says Saved

Use Jevitate to investigate a misleading success message, inspect browser evidence, and recheck a fix against the failure that prompted it.

014

Before hiring more engineers, find where delivery gets stuck

Trace one recent feature through your team to distinguish a capacity shortage from coordination, architecture, and product-decision bottlenecks.