Jev in production › Code and security review

prompt-fossils

CLI audits CLAUDE.md and AGENTS.md files; Jev's judgment narrowed 617 pattern hits to 460 real findings across 101 of GitHub's 1,000 most-starred repos (author).

Open on GitHub ↗

460 of 617measured, as published by the source
Use
Code and security review
Industry
Developer tools
Form
Open-source tool
Stage
Announced
Listed
2026-09-25
Found via
github
Repository
stas4000/prompt-fossils
Stars
2
Forks
1
Last push
2026-09-23
Language
HTML
License
MIT
prompt-fossils screenshot
docs/preview.png in the stas4000/prompt-fossils README, MIT; shown from GitHub.

The README opens with

Your CLAUDE.md, AGENTS.md and skills were written for older models. prompt-fossils finds the lines that now hobble Claude Opus 5.5 (the shouting, the bare "never" lists, the step scripts, the "think step by step") and tells them apart from the lines that are load-bearing.

One file, Python 3.10+, standard library only. Built with Claude Opus 5.5 in Claude Code.

The 1,000 most-starred public repos were checked for a root CLAUDE.md or AGENTS.md on 23 September 2026.

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for code and security review