Agentic coding consultancy
Send me the repository and the question. I read what is actually there and tell you what is wrong, what is fine, and what will cost you in six months.
Architecture calls, pull request review, and deciding what a coding agent is allowed to touch make up most of my week as a senior frontend engineer at Bejamas. Here, you book that same category of review independently, by the hour, against one decision on your own project.
Custom quote
- A review of your existing content schema for gaps an agent could exploit or get wrong.
- An audit of your AGENTS.md or equivalent instruction file against what the repository actually contains.
- A Core Web Vitals and rendering-performance pass on a specific page or template, with fixes ordered by impact.
Experience
The kind of review I run before an agent gets wider write access to a codebase, scoped down to your site.
FAQ
Questions, answered
What does this cover?
Architecture decisions you're not confident in, a pull request or code review you want a second opinion on, a Core Web Vitals regression nobody's traced yet, and - if you're already running an AI-editable setup - whether your content schema and CI actually catch what they should, and whether your AGENTS.md still matches what the repository contains. The number of AGENTS.md files that describe a structure the codebase abandoned six months ago is the single most common finding in that last category.
The other half is onboarding: the rules, commands and project docs that let a new developer, or Claude Code and Codex, open the repository and ship a correct first change without anyone walking them through it. I set that up on client codebases before the team grows, and I write it for the agent as its actual reader, not as a README nobody opens.
What does this not cover?
Building the guardrails from scratch. If you don't have a schema or CI validation yet, that's AI-editable websites, no CMS, scoped as a fixed-price build.
In practice
How a code review runs
Send me the repo and the question
What's about to change: an architecture decision, a stalled Core Web Vitals score, a pull request you're not sure about, or an agent getting wider write access.
I read the code before I quote
I clone the repository, go through the part your question touches, and only then give you a fixed hour estimate. No retainer running in the background.
You get the review in writing
Findings ordered by what bites first, and the small ones opened as pull requests rather than described in a document you then have to implement.
Agentic coding consultancy
Custom quote
- A review of your existing content schema for gaps an agent could exploit or get wrong.
- A check on whether CI actually fails on the cases it should.
- An audit of your AGENTS.md or equivalent instruction file against what the repository actually contains.
- The rules, commands and project docs a coding agent needs, written so a new developer or an agent ships a first change without a walkthrough from you.
- A Core Web Vitals and rendering-performance pass on a specific page or template, with fixes ordered by impact.
- A code review on a pull request or an architecture decision you're not confident in, before it ships.
Get in touch