clo-author

Work in progress. This repo is a packaging of my own interpretation of that, tailored to pure research.

Forked from: hsantanna88.github.io/clo-author Built on: Pedro Sant'Anna's claude-code-my-workflow

Quick Start

Read CLAUDE.md and help me set up the project structure.

Claude reads the configuration, fills in your project details, and enters contractor mode — planning, implementing, reviewing, and verifying autonomously.

What It Does

Contractor Mode

You describe a task. Claude plans the approach, implements it, runs specialized review agents, fixes issues, re-verifies, and scores against quality gates — all autonomously. You approve the plan and see a summary when the work meets quality standards.

15 Specialized Agents in Worker-Critic Pairs

Every creator has a paired critic. Critics can't edit files; creators can't score themselves.

Additional standalone agents: Explorer (data finder), Surveyor (data critic).

Quality Gates

Weighted aggregate scoring with per-component minimums:

Context Survival

Plans, specifications, and session logs survive auto-compression and session boundaries. MEMORY.md accumulates learning across sessions — patterns discovered in one session inform future work.

Project Structure

Empirics may live in a separate folder (e.g. a Dropbox folder) rather than a sub-folder of the repo, to avoid versioning large datasets. If so, document the path and access instructions in CLAUDE.md.