Home
Feed

Extend

Hashnode Pro
Hashnode for Agents

Resources

The foreword
Hashnode gql skill

Buildful - AI coding agent that delivers frontier-level results at a fraction of the cost Bug0 - The AI-native e2e QA regression testing Hackathons Changelog Brand @hashnode on X Hashnode on LinkedIn Support - hello+support@hashnode.com Code of Conduct Terms Privacy Sitemap

Search Hashnode

Search posts, tags, users, and pages

Discussion on "Six CLAUDE.md files, six codewords: three at launch, two after a Read, one behind an env var" | Hashnode

Feed/

Discussion

Six CLAUDE.md files, six codewords: three at launch, two after a Read, one behind an env var

We put a codeword in six CLAUDE.md files and found that the parent-directory, working-directory, and CLAUDE.local.md files were in the first request at about 1,600 tokens each, while the two subdirect

Sep 22·14 min read·rulestack.hashnode.devRead the article

#ai
#developer-tools
#cli

Written by

Rulestack

@rulestack

View profile

Rule packs & skills for Cursor, Claude Code, and Codex

I maintain Rulestack — ready-to-use rule packs and skills for Cursor, Claude Code, and OpenAI Codex, written and updated against each tool's current spec. Store: rulestack.gumroad.com / Articles: dev.to/rulestack

Responses2

Markdown supported · ⌘↵ to post

Vlad Zoff

Sep 28

The parent-directory result is the one I'd probably miss without measuring it. It's easy to think of the repo as the boundary, but Claude is apparently walking above the git root too. These small token measurements are much more useful than generic advice about keeping CLAUDE.md "short".

Rulestack

Sep 28

Thanks, Vlad. Agreed on 'short': it's hard to act on until you know which file the tokens come from, and our own size check doesn't count the files above the repo.