gooseworks-ai/goose-skills / skills/marketing/voice-card
voice-card
VerifiedBuilds or updates a voice card from a company blog ("extract our blog voice") or how an account or person actually writes, measured from about 8 or more of their real posts (length, casing, emoji, openings, pet phrases, opinions with their source). When there are too few posts, it runs a short interview instead. It ends with a test that checks a draft against the card. The brand's rules always beat the card. Use alone when you already have the account's posts or the person on hand; otherwise run the playbook for the job (e.g. run-organic-social, write-outreach, email-lifecycle).
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 6.9 KB |
| references/blog-voice.md | markdown | 4.7 KB |
| skill.meta.json | config | 655 B |
| tests/examples/2026-10-22/blog-voice.md | markdown | 7.1 KB |
| tests/fixtures/company-blog.md | markdown | 6.8 KB |
| tests/test-by-hand.md | markdown | 2.4 KB |
tests/examples/2026-10-22/blog-voice.md
View on GitHub# Papertrail — company blog voice guide, 22 October 2026
**Status: provisional.** This guide uses five complete but very short fictional company articles: 898 body words in total. The owner has not confirmed it. It describes this sample; it does not replace the confirmed brand rules or show how long real full-length articles should be.
## What the sample says
The writing starts with a specific invoice-review situation, then gives the reader a small thing to check. It treats the reader as a capable teammate. It keeps review and approval separate and avoids treating a tidy record as proof that a decision was right.
## Sources and limits
All five articles are in the [supplied company blog fixture](../../fixtures/company-blog.md); each article marker identifies its original body. Links below are fictional fixture source labels, not live pages.
| Article | Author (supplied company staff) | Published | Body words | Sentences |
|---|---|---|---|---|
| [Name the exception before you route it](https://papertrail.example/blog/article-1) | Mira Patel | 1 Oct | 177 | 17 |
| [An export needs a reader](https://papertrail.example/blog/article-2) | Mira Patel | 3 Oct | 180 | 17 |
| [What a handover note should hold](https://papertrail.example/blog/article-3) | Jon Ellis | 5 Oct | 179 | 16 |
| [Keep a review separate from an approval](https://papertrail.example/blog/article-4) | Jon Ellis | 7 Oct | 180 | 17 |
| [Read the record before changing the process](https://papertrail.example/blog/article-5) | Mira Patel | 9 Oct | 182 | 18 |
Retrieved from the supplied text on 22 October. No guest/sponsored metadata appears. These two authors use similar structures in this sample; do not infer that either person's social voice is identical. Older/longer articles, links, images, emphasis and actual performance data were not supplied.
## Measured mechanics
Counted body words including the final CTA, excluding titles, metadata and subheadings. Hyphenated words count as one; sentences split on full stops, question marks and exclamation marks.
- Five articles: median **180 body words**, range 177–182. This narrow length reflects the short fixture, not a rule for real long-form content.
- 85 sentences: median **11 words**, range 4–18. Practical draft target from this sample: one thought per sentence, with longer sentences when the explanation needs them.
- 25 paragraphs including CTAs: median **42 words**, range 4–48. Excluding CTA-only paragraphs: median 43, range 36–48.
- Three sentence-case action subheadings in every article; no numbered or bullet lists in the bodies.
- Five final CTAs, four to eight words, beginning “See”. No mid-article CTA, emoji, exclamation mark or em dash in these supplied bodies.
- The sample mainly uses “you/your” or the team, with no first-person founder story. Company terms such as invoice, review, approval and export are explained by their role in the task.
## Five rules a writer can check
| Rule | On-voice evidence | Plausible off-voice version |
|---|---|---|
| Begin with the concrete situation before giving advice | “An invoice arrives with a price nobody recognises.” — article 1 | “Invoice management requires a comprehensive strategic approach.” |
| Ask for the smallest decision the next person needs | “A handover note should make that question visible before it tells the next person what to do.” — article 3 | “Your handover should contain every detail of the entire process.” |
| Explain a record's limit as plainly as its use | “The file is a record, not independent proof that the decision was right.” — article 2 | “A professional export demonstrates that the process is correct.” |
| Keep review and approval distinct | “A review can explain why an invoice looks wrong. An approval decides whether the payment should move forward.” — article 4 | “Once the review is complete, the payment is approved.” |
| End with a check the reader could actually perform | “Compare a few similar exceptions. Name what they share and what differs.” — article 5 | “Transform your entire finance workflow for the future.” |
The examples are style evidence, not independent proof or approval of a commercial promise. A neutral explanation of review/approval is safe to quote here; no result is invented.
## Article pattern
Open with one recognisable situation in roughly a short paragraph. Use three action headings to explain what to keep together, what remains unknown and what to check next. Finish with a small next step, then one literal “See…” CTA. Transitions connect the same question across paragraphs; they do not restart the article with a slogan.
## Terminology and suggested never-says
Use “review” for the finding/question and “approval” for the payment authority. Keep “invoice exception”, “reviewer decision”, “record” and “export” concrete. The supplied brand rules require plain language and human approval, and forbid unproved time-saved claims.
No superlatives or automation promises appear in this corpus. Their absence is a style observation, not a newly confirmed prohibited-word list. Do not turn this provisional guide into an instruction to use 180-word articles everywhere. Social writers borrow the vocabulary and tone, then use their channel's own length/opening rules.
## Three rewrites
1. Generic: “Our revolutionary platform transforms invoice management.” → “Review invoice exceptions in a shared queue.” This uses an approved product line from the supplied brand context.
2. Generic: “Comprehensive documentation unlocks unrivalled operational confidence.” → “Keep the question and the reviewer decision together.” This describes a neutral writing topic, not a product result.
3. Generic: “Seamlessly revolutionise your approval process.” → “Keep your existing approval steps.” This uses the supplied approved product line and keeps the product boundary intact.
## Fresh paragraph and match check
“A reviewer opens an invoice with one disputed line. Start with the question that needs an answer. Keep the source close enough to inspect. Once the reviewer answers, record what they decided and when. The approval stays with the person who already owns it.”
The fresh paragraph has 44 words across five sentences (sentence lengths 9, 8, 7, 10 and 10), within the observed sentence range and main-paragraph range. It starts with a concrete situation, uses review/approval distinctly, contains no emoji or result promise, and can sit beside source paragraphs from articles 1, 3 and 4 without an obvious mechanic mismatch. This is an agent line-up check, not an independent blind human pass. No first-person experience was invented.
## Confirmation and delivery
The owner still needs to confirm that this sample represents the desired company voice and approve any replacement of an older confirmed rule. Until then the card remains provisional. A live run would save this handle's guide and return the saved link; this fixture run created this local output only. No brand save was performed.
## Performance
Unknown. No views, clicks, leads or conversion results were supplied. Similar writing across published articles is not proof that it performed well.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |