Claude Code's real unlock is repo structure: CLAUDE.md, skills, hooks and progressive docs
ifioknkem · x · 2026-09-08
Developer ifioknkem argues the real unlock for Claude Code isn't better prompts but structuring your repository. His "anatomy of a Claude Code project":
- CLAUDE.md as repo memory: keep it short — purpose, repo map, rules and commands; overly long files cause the model to miss critical signals.
- .claude/skills/: turn recurring workflows (code review checklists, refactoring playbooks, debugging, release procedures) into reusable expert modes.
- .claude/hooks/ as guardrails: models forget, hooks don't — auto-format after edits, run tests after core changes, block sensitive directories (auth, billing, migrations).
- docs/ as progressive context: don't stuff prompts; let Claude navigate architecture docs, ADRs and runbooks.
- Local CLAUDE.md in critical modules: e.g. src/auth/, src/persistence/, infra/ — dramatically cuts mistakes in danger zones.
His takeaway: prompting is temporary, structure is permanent — a well-structured repo turns AI from a chatbot into a project-native engineer.
More from coding & agent
- AI-generated PCBs plus desktop fabrication: from scrap board to working circuit in an hour — debreuil · 2026-09-08
- AI model Astra wins 40 of 43 AtCoder contests; Opus 5 dip hints at training contamination — teortaxesTex · 2026-09-08
- Open-source BYOK playground forces LLMs to output legal moves across five board games — wilsonye · 2026-09-08
- Astra Transcribes a Piano Part From YouTube, Then Builds a Playable App — danshipper · 2026-09-08
- Prompt Injection Attacks on AI Agents Up 340% in 2026; Fixes Must Be Architectural, Not Prompting — Thionne_WTZ · 2026-09-08
- Dev proposes open protocol for agent tool discovery at $0.05 per request — kleffew94 · 2026-09-08