Skip to content
SkillScript-led1,299 tokens in use

Preservation Verifier

Compares an original and a rewritten text to confirm code, quotes, tables, links, numbers and headings survived, and returns pass, review or fail.

5,770 stars on the repository MIT Licenseconorbronsdon/avoid-ai-writing · skills/preservation-verifier
/plugin install avoid-ai-writing@conorbronsdon-skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Preservation Verifier costs

54tokens idle
1,299tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Script-led

Script-ledMixedModel-led

Most of this skill is scripts, so it behaves much the same whichever model runs it.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: scripts/patterns.js, scripts/validate.js.

Touches

shelldetectednetworkdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter15/15
  • Structure9/25
  • Examples12/15
  • Output15/15
  • Size10/15
  • Resources5/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Fair
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 59
  • Hidden characters−20

    characters that render as nothing or reorder the text

    scripts/patterns.js, line 72
  • Hidden characters−20

    characters that render as nothing or reorder the text

    scripts/patterns.js, line 73

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity96
  • Maintenance100
  • Community60
  • Maturity32
  • Momentum40

Preservation Verifier checks that a rewrite or file edit kept what the Avoid AI Writing rulebook says to protect. It needs both versions and runs a bundled Node validator with a residual-pattern policy that keeps mechanical errors blocking and reports pattern growth as a warning.

It adds a separate semantic review for meaning, attribution, quantities, negation, conditions and uncertainty, and returns PASS, REVIEW or FAIL with the repair owner. It never rewrites content itself.

What You Get

  • A PASS, REVIEW or FAIL result for a rewrite
  • Blocking preservation errors and warnings from the validator
  • A separate semantic review of meaning, attribution, numbers, negation, conditions and uncertainty
  • The right repair owner on failure: the rewriter for returned text, the file editor for a named file
  • A statement of whether the single allowed repair has already been used

Common Use Cases

  • Checking that a rewritten article kept its code blocks, quotes and links
  • Confirming a file edit did not change numbers, headings or frontmatter
  • Reviewing a rewrite for growth in AI patterns after cleanup
  • Validating a rewrite of a prompt that describes people, so identity details survive
  • Stopping a workflow when a second check still fails

What Makes It Unique

It keeps mechanical checks and semantic judgment apart and says which one produced each result, so a model-only review is never presented as validator output. The repair loop is capped at one re-entry, and a failure counts as a blocking result: a rewrite is not complete just because text was produced.

Under The Hood

The skill loads one instruction file and bundles a Node validator with its pattern engine. The agent runs the validator in the shell on two text files, which only reads them and prints results; it can also be called as a JavaScript function. The result is reported by the agent and nothing is written to the files. It makes no network calls, and falls back to a manual comparison when Node cannot run.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers