Skill collection · established

blader/humanizer

Portable writing skill that identifies and removes common signs of formulaic AI-generated prose while preserving meaning, facts, and quoted material.

writingeditingagent skills
Installation GuideInstruction Manual

What it contains

One detailed Agent Skill with 35 documented pattern families, before-and-after examples, voice-matching guidance, packaging checks, and install paths for many agents.

Typical uses

Edit a draft so it sounds more specific, natural, varied, and consistent with the writer's own voice.

Why people choose it

Choose it for a focused revision pass after the facts and structure are already correct, especially when a draft is repetitive, over-signposted, or padded.

License
MIT
Maintenance
active
Latest release
See GitHub Releases
Last meaningful update
2026-08-18
Maturity
established
Production readiness
production

Classification

Subject domains

writingeditingagent skills

Task categories

style revisionvoice matchingAI-pattern cleanup

Project phases

draftingeditingreview

Secondary types

plugin

Select when

  • A draft contains repetitive AI-writing patterns
  • Need editing that preserves meaning and quoted text
  • Want a portable skill across several agent hosts

Do not select when

  • Need fact-checking or source research
  • Need to conceal prohibited AI use or misrepresent authorship
  • The source text must remain verbatim
Implementation complexity
low
Setup effort
low
Learning curve
low
Confidence
high

Routing heuristic 100.0%

Routing specificity
5/5
Setup simplicity
5/5
Operational maturity
5/5
Composability
5/5
Local control
5/5
Security boundary clarity
5/5
First-party routing documentation
5/5

Strengths

  • Narrow, understandable scope
  • Extensive pattern examples
  • Global, project, plugin, and desktop installation options

Poor-fit scenarios

  • Need fact-checking or source research
  • Need to conceal prohibited AI use or misrepresent authorship
  • The source text must remain verbatim

Limitations

  • Style judgments remain subjective
  • A human review is still required
  • It does not verify facts or citations
humanizer skill
skills/humanizer/SKILL.md

Review prose against 35 documented AI-writing patterns and revise it for naturalness, specificity, rhythm, and voice while protecting meaning and quotations.

editingvoice
Languages
Markdown; Python only for package-validation scripts
Frameworks
Agent Skills specification, Claude plugin
Runtimes
Compatible agent host; Node.js for npx skills installation
Operating systems
macOS, Linux, Windows through the selected host
Installation methods
npx skills, Claude plugin, Claude Desktop ZIP, or manual clone
Interfaces
Agent Skill and Claude slash command
Required credentials
None
External services
Only the chosen model/agent provider
Hardware
No special hardware
Major dependencies
Compatible agent host

Compatibility notes

Apple silicon and Tahoe require no special runtime beyond the selected agent. For Claude Desktop, use the release ZIP rather than GitHub's source archive because the desktop loader rejects the repository symlink layout.

One-sentence semantic summary

Portable writing skill that identifies and removes common signs of formulaic AI-generated prose while preserving meaning, facts, and quoted material.

Capability keywords

humanizerwritingeditingAI prosevoicestyle

User intent phrases

  • Make a draft sound less formulaic and more like a person
  • Match an existing writing voice

Negative match phrases

  • Need source verification
  • Need verbatim preservation

Differentiators

A single focused skill with a large, explicit taxonomy of writing patterns and safeguards for quoted text.

Security considerations

The skill edits text through the host agent; review the diff and do not use it to evade disclosure, academic, or platform rules.

Privacy considerations

Drafts are handled according to the selected agent provider. Do not submit confidential writing to a service whose retention policy is unacceptable.

Uncertainty

Unresolved items

None documented.

Inference notes

Routing guidance is an analyst synthesis based on the first-party repository and README.

Evidence

  • repository_metadata · Repository identity and current GitHub metadata checked on August 18, 2026.
  • readme · Primary first-party README inspected.

Repository usage guidance

installation summary
Install globally or per project with npx skills, use the Claude plugin, or upload the release ZIP to Claude Desktop
basic usage summary
Provide the draft and, when possible, a sample of the intended voice; then review every change
documented entry points
humanizer SKILL.md and /humanizer:humanizer
key configuration files
Agent skill registration
important directories
skills/humanizer/
documentation paths
README.md and SKILL.md
example paths
README.md pattern examples

Relationships

complementary
high
Marketing Skills can create campaign copy and Humanizer can perform a separate final style pass after factual and strategic review.

Recommended combinations

Draft with a relevant Marketing Skills component, verify claims and brand voice, then run Humanizer as the last reversible editing pass.

Routing rules

rule_052: Need to remove formulaic AI-writing patterns from a draft P948

Rationale

Portable writing skill that identifies and removes common signs of formulaic AI-generated prose while preserving meaning, facts, and quoted material.

Required conditions

A draft contains repetitive AI-writing patterns

Preferred

blader/humanizer

Fallback

Choose the closest narrowly scoped existing repository only after comparing permissions and workflow ownership.