Open
Conversation
Content strategy review for codebases where the words ARE the interface: chatbots, notifications, onboarding flows, error messages, CLI output. Extracts every user-facing string, evaluates it against a voice guide (or infers one), and produces a string audit with concrete rewrites. Draws on the public style guides of Mailchimp, Microsoft, Apple HIG, Shopify Polaris, Google Material, and GOV.UK. Encodes what content strategy teams actually check for: - Voice vs tone (Mailchimp): voice is constant, tone shifts by moment. - Plain English, active voice, verb-first imperatives (GOV.UK, Shopify). - Scan test: first 5 words must carry the point (Google, Microsoft). - The 2am test for safety-critical strings. - Weigh every word (Shopify, Microsoft). - Anti-patterns: clinical, corporate, robotic, surveillance, generic AI. Phases: 1. Understand the voice (find or propose a voice guide, save as VOICE.md) 2. Extract strings and classify by emotional stakes 3. Audit against the voice guide 4. Optional i18n readiness check (back-translation round-trip using the LLM's multilingual capability; user can punt for English-only) 5. Rewrite and emit a report Use when asked to "review the copy", "content review", "voice audit", "string review", "check the messaging", or "tone check".
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds
/content-review, a content strategy review skill for codebases where the words ARE the interface... chatbots, notifications, onboarding, error messages, CLI output. Extracts every user-facing string, evaluates against a voice guide, and produces a string audit with concrete rewrites.Why
Most "code review" skills look at logic, types, structure. But for conversational products, the strings are the product. A misplaced enum value or a cheerfully-exclamation-pointed error message can be the whole UX failure, and traditional review misses it.
What it encodes
Draws on the public style guides of Mailchimp, Microsoft, Apple HIG, Shopify Polaris, Google Material, and GOV.UK. Key principles:
Phases
Lightly Battle-tested
Ran on a real WhatsApp chatbot codebase (aging-parent companion product). Found real bugs: enum values leaking into strings ("Mom's doing good" from
mood_signal=good), scan test failures on a distress alert, hardcoded "Mom" where a{parent_name}template should have been.Test plan
bun test test/gen-skill-docs.test.ts— 350 pass, 0 fail--host all