One operating contract replaces duplicated workflow
The candidate brings source fidelity, voice matching, invocation modes, the rewrite loop, and delivery rules into one path instead of repeating the process at the top and bottom of the Skill.
Humanizer Skill · Source-faithful editing
Make prose sound human without losing what the source says.
The optimized Humanizer inventories claims, protected literals, voice, and pattern clusters before it rewrites, audits, and converges. The canonical 33-pattern catalog remains part of the package.
Humanizer is a text-editing Skill that removes clustered AI-writing patterns while preserving source information, register, and authorial voice.
The candidate brings source fidelity, voice matching, invocation modes, the rewrite loop, and delivery rules into one path instead of repeating the process at the top and bottom of the Skill.
Every paragraph, claim, protected literal, register, and relevant pattern cluster is accounted for before the rewrite is delivered.
The three invocation modes now state what the caller receives, while the package keeps the original patterns, references, and install identity.
Keeps names, numbers, dates, quotes, citations, links, and requested register visible during the rewrite loop.
Treats the preserved 33-pattern catalog as evidence to inspect in context rather than flattening isolated human signals.
Repeats the audit until every identified cluster is addressed and every source claim is represented.
A supplied writing sample outranks default style rules when the task asks for a voice match.
Run a source-faithful final edit over text that feels generic, robotic, or over-polished.
Supply a writing sample and keep its rhythm and register while removing clustered patterns.
Preserve frontmatter, data, code blocks, and link targets while revising prose in place.
Verified source-diff evidence
This release proves canonical source lineage, package inventory, the optimized true diff, and an immutable artifact. It does not claim runtime outcomes or visual comparisons.
.scratch/kan-300/production-release/source-evidence/candidates/humanizer/original-inventory.json#/entries?sha256=ce13a6a18e81ee0e3f319ae50a57e539353bf7a3cead91217264b5072bb1ee63.scratch/kan-300/production-release/source-evidence/candidates/humanizer/optimized-inventory.json#/entries?sha256=3e3540110e06265af996004f11190357933e974bff2d7373d1557c9ad818294d.scratch/kan-300/production-release/source-evidence/candidates/humanizer/change-set.json#/evidenceUnits/0?sha256=5d8d4d2fbe50420fc83ee954fd33023a5571cd517ad5cb4d6194b861517fcf4d.scratch/kan-300/production-release/source-evidence/candidates/humanizer/change-set.json#/evidenceUnits/1?sha256=5d8d4d2fbe50420fc83ee954fd33023a5571cd517ad5cb4d6194b861517fcf4d.scratch/kan-300/production-release/source-evidence/candidates/humanizer/change-set.json#/evidenceUnits/2?sha256=5d8d4d2fbe50420fc83ee954fd33023a5571cd517ad5cb4d6194b861517fcf4d.scratch/kan-300/production-release/source-evidence/candidates/humanizer/change-set.json?sha256=5d8d4d2fbe50420fc83ee954fd33023a5571cd517ad5cb4d6194b861517fcf4d.scratch/kan-300/production-release/source-evidence/candidates/humanizer/change-set.json?sha256=5d8d4d2fbe50420fc83ee954fd33023a5571cd517ad5cb4d6194b861517fcf4d# Humanizer This edition derives from blader/humanizer v2.9.1 at 523374dee72d67c7b2b5f858ea0094ffda49c3ac. ## Runtime contract Inventory the full input, protect its facts and literals, rewrite against the intended voice, audit the result, and converge before delivery. Pasted-text, file, and embedded invocation modes define their own output boundary. The package keeps the canonical 33-pattern catalog, examples, false-positive guidance, voice calibration, MIT license, and nine-file natural shape. This release edition does not include a matched output artifact.
Use these phrases to activate this skill in your AI coding assistant: