blader/humanizer

Agent skill that removes signs of AI-generated writing from text

View on GitHub ↗Jump to charts ↓

Summary Information

Updated 8 minutes ago
Added to GitGenius on August 31st, 2026
Created on January 18th, 2026
Open Issues & Pull Requests: 24 (+0)
GitHub issues: Enabled
Number of forks: 3,429
Total Stargazers: 39,261 (+1)
Total Subscribers: 220 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 23.0 days
Mean response time: 47.2 days
90th percentile: 171.7 days
Tracked items: 95

How this project is maintained

Around half of the issues opened in the past year never receive a reply. Only 15% of issues opened in the past year have been closed. Three people close 92% of everything that gets resolved.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 11
New in 7 days: 3
Closed in 7 days: 0
Avg open age: 32 days
Stale 30+ days: 1
Stale 90+ days: 0

Recent activity

Opened in 7 days: 2
Closed in 7 days: 0
Comments in 7 days: 0
Events in 7 days: 0

Top labels

No label distribution available yet.

Detailed Description

Humanizer is an agent skill that rewrites AI-generated text to sound naturally human-written while preserving factual accuracy and meaning.

The tool addresses the problem that large language models tend to produce statistically likely prose that exhibits recognizable patterns—inflated language, overused phrases, awkward constructions, and vague sourcing. Humanizer works by applying 35 patterns derived from Wikipedia's documented signs of AI writing. It makes an initial rewrite without preserving the original structure, then checks the result against those patterns and verifies that factual claims remain grounded in the source material or the writer's own knowledge. The tool shows its work, displaying both the intermediate rewrite and a critique of remaining artificial elements before delivering the final version.

Humanizer suits anyone using AI writing assistants who wants output that reads naturally without sacrificing accuracy. It works with any agent that supports skills and handles mixed-content files by rewriting only prose while leaving code, data, frontmatter, and link targets untouched. The tool can match a writer's voice by following a provided writing sample, adapting its style rules to the sample's rhythm, word choice, punctuation, and deliberate quirks rather than applying defaults. For technical and reference writing, it maintains neutral, plain language.

The project shows consistent maintenance with regular updates addressing the pattern library and tool functionality. Development activity includes ongoing refinement of the detection and rewriting logic to handle edge cases in how AI patterns manifest across different writing contexts. The tool receives periodic improvements to its handling of mixed-content files and its ability to preserve writer intent while removing artificial markers.