Skip to content

Add optional Positive Voice Guide section#99

Open
adelaidasofia wants to merge 1 commit intoblader:mainfrom
adelaidasofia:feat/positive-voice-guide
Open

Add optional Positive Voice Guide section#99
adelaidasofia wants to merge 1 commit intoblader:mainfrom
adelaidasofia:feat/positive-voice-guide

Conversation

@adelaidasofia
Copy link
Copy Markdown

What

Adds a new optional section, Positive Voice Guide, right after the existing Voice Calibration (Optional) section in SKILL.md. +20 lines, no deletions.

Why

Voice Calibration teaches voice by example (a writing sample). I've found in heavy use that examples alone leave gaps: humanizer strips AI tells well but the result still lands generic-human, not the writer's specific human. Especially when the prose is right for one surface but wrong for another (e.g. journal vocabulary on a business surface), a sample doesn't catch it.

A positive voice guide teaches voice by description: a small file the writer maintains with their positive principles, hook patterns, audience routing (different vocabulary for different surfaces), and concrete anti-examples. That gives the rewrite a North Star after the AI tells are stripped.

The two are complementary, not competing. Sample = pattern data. Guide = explicit rules. Both feed the rewrite.

How

Optional and self-contained:

  • If the user provides a guide (inline, file path, or convention like voice-style.md at project root), humanizer reads it before rewriting and uses it as a quality check on each paragraph.
  • If no guide is provided, humanizer falls back to Voice Calibration (if a sample exists) or PERSONALITY AND SOUL defaults, exactly as today.

No existing behavior changes for users who don't have a voice guide.

Notes

  • Mirrors the structure of the existing Voice Calibration (Optional) section so it reads as a sibling, not a replacement.
  • I've been running this pattern in my own fork against my Obsidian vault for a few weeks. The measurable improvement is on third-round drafts: rewrites that were "clean but generic" now match the surface (LinkedIn vocab on LinkedIn, essay vocab on essays, etc.) on the first pass.
  • Happy to revise the wording or trim the section if it feels too long for SKILL.md. The core idea is small; everything past the first paragraph is just usage detail.

Voice Calibration teaches voice by example (a writing sample). This
adds a complementary mechanism: voice by description (a guide that
states the writer's positive principles, hook patterns, audience
routing, and concrete anti-examples).

Without a positive guide, even a clean rewrite can land flat:
generic-human, not the writer's specific human. The guide acts as
the rewrite's North Star after AI tells are stripped.

Optional and self-contained. If no guide is provided, the existing
Voice Calibration / PERSONALITY AND SOUL fallbacks run unchanged.
spencergoldade added a commit to spencergoldade/humanizer that referenced this pull request Apr 22, 2026
Upstream blader/humanizer PR blader#99 (adelaidasofia). Pure additive.
Inserts between Voice Calibration and PERSONALITY AND SOUL. Gives the
rewrite a pull-toward anchor (explicit voice rules the writer maintains)
on top of the existing voice-calibration sample mechanism. Author
reports the measurable win is on "clean but generic" rewrites.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant