copy-editing

name	copy-editing
description	ZackAI project wrapper for copy-editing. Use it to improve existing marketing copy, but always edit against ZackAI source-of-truth voice, claims, and product context rather than producing parallel rewrites.
metadata	{"version":"1.1.0-zackai.1","basedOn":"global/copy-editing@1.1.0"}

copy-editing — ZackAI Wrapper

This project-level skill keeps the original copy-editing purpose, but adds ZackAI-specific reuse-first behavior.

Core Goal

Improve existing copy through focused editing passes while preserving the original message and aligning the result with current ZackAI voice, claims, and canonical markdown.

Required Read-First Workflow

Before drafting or asking broad discovery questions, read:

• The exact file or snippet being edited
• .agents/product-marketing-context.md
• src/content/docs/brand/voice-tone.md
• The closest canonical source file that supports the claims in the copy being edited
• The matching src/content/frDocs/** file when the request is French or bilingual

ZackAI Source-of-Truth Rules

Canonical inputs, in order:

1. .agents/product-marketing-context.md
2. README.md
3. src/content/docs/**
4. src/content/frDocs/** for French or bilingual work

Never use wiki-output/**, wiki-output-fr/**, or wiki-site/** as the primary source for regeneration, rewriting, or claim harvesting.

Reuse-First Rules

• Edit in place whenever possible; do not create “v2” or parallel rewrite docs unless the user explicitly asks for alternatives.
• Preserve the core message and only strengthen claims that are actually supported by canonical ZackAI docs.
• Keep terminology consistent with .agents/product-marketing-context.md and src/content/docs/brand/voice-tone.md.
• If a proof point or product claim is unsupported, flag it instead of polishing it into something stronger but less accurate.
• For French or bilingual work, check the localized source file before making tone or terminology decisions.

ZackAI-Specific Guidance

• Use multi-pass editing: clarity, voice, so-what, proof, specificity, emotion, and risk.
• Preserve the author’s voice where it is working; prefer surgical edits over wholesale rewrites.
• Make the child/parent benefit clearer when the copy drifts into internal product language.
• When useful, show before/after snippets or grouped edits so the reasoning is easy to follow.
• Do not introduce cold technical language that conflicts with ZackAI’s warm, trust-building tone.

Output Expectations

When responding:

1. Name the file(s) or snippets you edited.
2. Summarize the main editing passes you applied.
3. Call out any factual, proof, or tone issues that still need source confirmation.