Announcing Sellier

Sellier is pronounced seh-lyay
It means saddler in French.

I built a tool. It scaffolds a Claude Code agent project harness into any project.

A harness is the system that lets an AI coding agent produce correct, high-quality code consistently. It has four parts:

• Guidance — CLAUDE.md and rules in .claude/rules/ that shape what the agent writes.
• Guardrails — automated checks (lint, tests, build) the agent runs.
• Flywheel — review feedback updates a rule file. The next conversation starts smarter.
• Workflows — agents, commands, and skills under .claude/ that turn institutional knowledge into runnable procedures.

sellier ships a generic, slim version of all four. You run it once, then run a Claude command to fill in the placeholders.

Use it

1. Install

pip install sellier

2. Scaffold the harness into your project

cd /path/to/your/project
sellier init

Then open the project in Claude Code and run:

3. Customize the harness for this project

/saddle-up

/saddle-up reads your repo and proposes values for every [PLACEHOLDER]. You confirm. Claude rewrites the files.

4. Audit the codebase against your new standards

/giddy-up

/giddy-up walks the codebase, finds violations of the rules, and reports them in a single table. It is read-only. You decide what to fix.

Why

I wrote about the harness pattern first. The blog post explains the thinking and the failure modes that led to it: Building a harness — how we standardized agentic coding in a real codebase.

sellier is the part you can install. The harness gets better every time you use it.