unslop-file

To remove AI tells from markdown files while preserving code blocks, URLs, paths, commands, and headings, run a script-driven pipeline that rewrites AI-style prose into human text. It executes a two-pass audit and creates a backup before overwriting.

Yes, you can humanize memory files like CLAUDE.md and AGENTS.md deterministically using a regex-based rewriting pass, which avoids LLM usage entirely while enforcing a safe backup and validation step.

Rewriting markdown files with Claude preserves YAML frontmatter, file paths, commands, and headings exactly as-is. It targets AI-style prose for rewriting while leaving technical details untouched during the two-pass audit.

The best way to batch process multiple markdown docs is using a script-driven pipeline that rewrites AI-style prose across multiple files. It creates a backup for each file and validates results to ensure technical details remain intact.

To revert a markdown file after AI text removal, use the generated backup file. The pipeline creates a backup before overwriting the original document, ensuring you can safely restore the exact prior state if needed.

You need Anthropic dependencies only for LLM-based rewrites via Claude. For deterministic humanization of memory files, the regex pass works independently, though Hugging Face and Torch dependencies power the broader script pipeline.