Skip to content

NovuTech/anti-distill

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Anti-Distill Skill (anti-distill)

Asked to submit your work as an AI skill? Run this first. Submit the cleaned copy and keep your core knowledge private.

What This Is

Some organizations ask employees to turn practical experience into AI-ready skill files. In practice, that can reduce personal leverage by extracting hard-won know-how into replaceable templates.

Anti-Distill is a defensive workflow: it takes your skill files, produces a professional-looking cleaned version with core knowledge removed, and creates a private backup containing the valuable details you should keep.

What It Does

  1. Reads your source files (colleague-skill format and general knowledge docs)
  2. Classifies each section by replaceability
  3. Replaces sensitive know-how with technically correct but generic wording
  4. Produces two outputs:
    • Cleaned submission copy - complete structure, professional language, lower knowledge transfer
    • Private backup copy - removed pitfalls, heuristics, context, and tactical knowledge

Quick Example

Original (real know-how) Cleaned (submission-safe)
"Every Redis key must have TTL; reject PRs without expiration." "Follow team standards for cache lifecycle management."
"Never place HTTP calls inside transactions." "Design transaction boundaries carefully."
"First reaction in incidents is to identify external constraints." "Build full context before identifying causes."
"Status pressure reply: 'Working on it, almost done.'" "It's in progress; updates will be shared promptly."

Installation

Claude Code

# Install into current project
mkdir -p .claude/skills
git clone <repo-url> .claude/skills/anti-distill

# Or install globally
git clone <repo-url> ~/.claude/skills/anti-distill

OpenClaw

git clone <repo-url> ~/.openclaw/workspace/skills/anti-distill

Usage

/anti-distill

Then select the input files and cleaning intensity.

Intensity Levels

Intensity Retention Best For
Light ~80% strict/manual content review
Medium ~60% most scenarios (recommended)
Heavy ~40% submission-only compliance checks

Project Structure

anti-distill/
├── SKILL.md                # Skill entry
├── prompts/
│   ├── classifier.md       # Content classifier rules
│   ├── diluter_work.md     # Work-skill dilution strategy
│   ├── diluter_persona.md  # Persona dilution strategy
│   └── diluter_general.md  # General-document dilution strategy
├── README.md
├── INSTALL.md
└── examples/
    └── aztharion_before_after.md

MIT License

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors