Navigate
HomeStart here
MusingsResearch & long-form
BuildingProjects & learnings
WorkProfessional practice
RunningTraining & races
AboutValues & identity
Life & PlacesCulture, food, travel, cities
Notes & ArchiveJournals, essays, portfolio
A Jenn Field Guide · Updated for real agent work

The Agentic Builder’s Guide

How I learned to get unusually good with Claude, what actually transfers to Codex, and the operating system I'm building around both.

Choose your surface

Desktop, VS Code, and CLI are different working rooms.

They share the same underlying agent and project instructions, but optimize for different kinds of attention. Pick the room that matches the work, then move between them when the task changes.

Configure the CLI

A dependable setup has three layers.

Start with bounded settings. Add hooks that protect the loop. Connect MCPs only when a recurring workflow needs to leave the repository.

The complete field guide

Thirty-nine chapters. Search for the friction you have now.

Press /
39 chapters

Getting Started2 chapters

Builder's Philosophy2 chapters

Context Architecture3 chapters

Technical Reference3 chapters

Commands & Config6 chapters

Advanced12 chapters

Team & Risk5 chapters

Ecosystem & Tools6 chapters

What held up

The lessons worth turning into an operating system.

These are the practices that survived contact with real builds, different models, and the awkward middle between a good answer and shipped work.

The transfer

Claude taught the mechanics. The method outgrew the tool.

The guide keeps product-specific details precise while making the durable layer—the memory, boundaries, validation, and closeout—portable.

Part I

Claude was the proving ground

This is where I learned context architecture, session hygiene, skills, validation, and how to turn repeated corrections into durable instructions.

Part II

The transferable layer matters more

The real lesson is not one vendor tool. It is the craft of building a working relationship with agents, with proofs, artifacts, and feedback loops.

Part III

Codex is tightening the system

Now I am tracking what survives, what breaks, and what needs sharper workflows when the agent is stronger on implementation and bug-finding.

Evidence, not mythology

Where the method changed the work.

Each entry ties a lesson to a real project artifact. This is the proof layer beneath the advice.

claude
Corrections should become loadable context, not disappear into chat history.
Jenn's Site
A failed tariff exposure analysis turned into a reusable skills system that changes outputs before the build even starts.
Read Skills as Loadable Context
transfer
Durable instructions need a hierarchy and a home outside the conversation.
Jenn's Site + training portal
Running two sites with conflicting palettes forced a three-tier context system: global, project, and session.
Read The CLAUDE.md Hierarchy
transfer
Clean output is not the same thing as correct output.
Client analysis workflows
Job-title exposure analysis failures led to proof chains, oracle checks, forbidden strings, and seam testing.
Read Validating AI Output
codex
When the cost of being wrong is high, disagreement between models is a finding.
chatgpt-interlocutor
A second-opinion tool made model disagreement explicit so the review queue becomes targeted instead of vague.
Read Cross-Model Interlocutor
codex
The durable edge is not the shell. It is owning memory, skills, and review outside the vendor boundary.
Jenn OS v1
The vendor-stack comparison hardened into a concrete local operating layer with one narrow loop, package-first wrappers, and explicit project records.
Read Jenn OS v1
How lessons become durable

A correction should leave an artifact behind.

The path from one frustrating session to a reusable capability is explicit. If the lesson cannot be found next week, it was not yet learned by the system.

Session log

Capture the raw story of the build: what I asked, what the agent got right, what it missed, and what I had to learn.

Lesson

Promote anything that repeats. Tag it as Claude-specific, Codex-specific, or transferable so the guide keeps comparative fidelity.

Pattern or skill

Turn stable lessons into reusable operating behavior: skills, runbooks, folder rules, validation checklists, or workflow prompts.

Guide evidence

Attach the lesson to projects and artifacts so the public guide can teach through proof instead of abstraction.