Goal

SkillDev tools

Read, criticize, and propose improvements for text

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Goal skill

What this skill tells your AI

The instructions your AI receives, as published by causify-ai/helpers in .claude/skills/text.criticize/SKILL.md and read by ahel’s review.

  • Review and criticize the text (markdown, blog posts, documentation)
  • Identify mistakes and improvement opportunities, focusing on factual errors, rather than style suggestions
  • Propose concrete improvements to content, flow, visual clarity, pacing, and audience fit

Workflow

Read the Material

  • Read the text passed by the user carefully

Identify and Rank Mistakes

  • Identify mistakes in the passed text

Definition

  • Criteria for mistakes:

    • Factual inaccuracy (wrong information, outdated claims)
    • Logical contradiction or inconsistency
    • Technical error (code doesn't work, command is wrong, API is misrepresented)
    • Broken reference (link is dead, file path is wrong, name is misspelled)
    • Only report issues you are absolutely certain about
  • Ranking severity:

    • High: Misleads reader or causes confusion
    • Medium: Could improve clarity or accuracy
    • Low: Minor inconsistency or incomplete statement

TODOs

  • Ignore the TODOs that already exist in the text

Suggest Improvements

  • Identify opportunities to improve clarity, structure, or completeness without changing facts

Definition

  • Types of improvements:

    • Clarity: Making text easier to understand
    • Completeness: Adding missing context or examples
    • Consistency: Aligning terminology or formatting
    • Readability: Breaking up dense sections or improving flow
  • Ranking severity:

    • High: Significantly improves comprehension or prevents misunderstanding
    • Medium: Moderately improves clarity or adds useful detail
    • Low: Minor improvement in readability or polish

Write the Results

  • The output format is a vim cfile following .claude/skills/cfile.rules.md

    <full path file>:<LINE_NUMBER>:1 [HIGH/MEDIUM/LOW] Short description of error
    <full path file>:<LINE_NUMBER>:1 [HIGH/MEDIUM/LOW] Why this is wrong and what the correct statement should be
    ...
    
  • Example

    msml610/lectures_source/README.md:12 [HIGH] Claims "pip install X" but pip package is named "X-lib"
    msml610/lectures_source/README.md:5 [HIGH] Add example of basic usage after the description
    helpers/docs.md:8 [MEDIUM] Link to API docs is outdated (2023 version, current is 2024)
    
  • Save this in a file related to the file being processed <DIR>/<FILE>.CRITICIZE.md

    • E.g., msml610/lectures_source/Lesson03.1-Knowledge_representation.txt -> msml610/lectures_source/Lesson03.1-Knowledge_representation.CRITICIZE.txt

Wait for User Approval

  • Wait for user to:
    • Select which mistakes and improvements using the index to fix
    • Provide any additional context or corrections
  • Once approved, implement the selected changes

Verification

  • Confirm the cfile follows .claude/skills/cfile.rules.md
  • Confirm every issue has a severity level and a file:line reference
  • Confirm the output file <DIR>/<FILE>.CRITICIZE.md was created

Signals

GitHub stars
145
Forks
160
Last commit
Sep 2026

ahel review

  • K1binfo
    installs-packages

Automated review, not a security audit. Ruleset v1+k2.

Advanced
Catalog kind
skill
Gateway key
text-criticize
Source
github.com/causify-ai/helpers