Writing and Editing Workflows
Overview
This umbrella covers prose work: drafting, editing, humanizing, voice matching, structure repair, and publication polish. The shared class is transforming text for a human audience while preserving meaning, facts, and the user's intended voice.
When to Use
- The user asks to rewrite, polish, humanize, de-AI, de-slop, or make text sound natural.
- The user provides a file or draft and wants edited prose.
- The task involves matching a user's writing sample or adapting tone for an audience.
- The output is a memo, article, email, post, documentation, or publication-facing text.
Shared Workflow
- Identify audience, format, tone, and whether facts need verification.
- Preserve meaning and constraints; do not invent claims or citations.
- Remove formulaic AI tells, weak structure, filler, and unsupported hype.
- Match user voice when a sample is provided.
- If editing a file, patch or write it and show the changed section/diff.
Labeled Subsections
Humanizing and de-AI editing
Use the absorbed Humanizer package notes for pattern lists, AI-writing tells, voice calibration, and final anti-AI passes.
Structural editing
Fix order, headings, paragraph purpose, and transitions before sentence-level polish.
Tone and voice matching
Extract concrete voice traits from samples: sentence length, punctuation, diction, paragraph openings, humor, and level of directness.
Package References
The Humanizer package is stored under references/absorbed/humanizer/, including its license file.
Verification Checklist
1---2name: writing-and-editing-workflows3description: Use when drafting, editing, humanizing, de-AI-ing, polishing, voice-matching, or preparing user-facing prose across docs, articles, emails, posts, memos, and publication text.4license: MIT5---67# Writing and Editing Workflows89## Overview1011This umbrella covers prose work: drafting, editing, humanizing, voice matching, structure repair, and publication polish. The shared class is transforming text for a human audience while preserving meaning, facts, and the user's intended voice.1213## When to Use1415- The user asks to rewrite, polish, humanize, de-AI, de-slop, or make text sound natural.16- The user provides a file or draft and wants edited prose.17- The task involves matching a user's writing sample or adapting tone for an audience.18- The output is a memo, article, email, post, documentation, or publication-facing text.1920## Shared Workflow21221. Identify audience, format, tone, and whether facts need verification.232. Preserve meaning and constraints; do not invent claims or citations.243. Remove formulaic AI tells, weak structure, filler, and unsupported hype.254. Match user voice when a sample is provided.265. If editing a file, patch or write it and show the changed section/diff.2728## Labeled Subsections2930### Humanizing and de-AI editing3132Use the absorbed Humanizer package notes for pattern lists, AI-writing tells, voice calibration, and final anti-AI passes.3334### Structural editing3536Fix order, headings, paragraph purpose, and transitions before sentence-level polish.3738### Tone and voice matching3940Extract concrete voice traits from samples: sentence length, punctuation, diction, paragraph openings, humor, and level of directness.4142## Package References4344The Humanizer package is stored under `references/absorbed/humanizer/`, including its license file.4546## Verification Checklist4748- [ ] Audience/tone/format identified.49- [ ] Meaning and factual claims preserved.50- [ ] AI tells and filler removed.51- [ ] Voice sample used if provided.52- [ ] Edited files read back or diffed when applicable.
quriosity-agent