tov-guidelines
Skill53 starsChanged 5 days agoMITView on GitHub
Extracts a company's tone of voice from its existing content (website, blog, social, docs) and turns it into usable guidelines: tone attributes with before/after, vocabulary, patterns and anti-patterns. Use when asked for tone of voice, brand voice or writing guidelines.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- matteotitta/repo-structure-agencytheir own version, first seen Nov 8, 2026
- matteotitta/claude-marketing-contexttheir own version, first seen Oct 15, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- matteotitta/claude-marketing-context★ 3same name
- matteotitta/repo-structure-agency★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- aeo-strategySkill
- battlecardsSkill
- case-studySkill
- company-contextSkill
- competitor-researchSkill
- icp-researchSkill
- linkedin-profileSkill
- positioningSkill
- product-launchSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 58
Save approved output as reference example: if user explicitly approves — see the premium reference.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
TOV guidelines
Extract actionable editorial tone of voice guidelines from a company's existing content. Two-phase workflow: Phase 1 analyzes website content to surface evidence-based voice patterns; Phase 2 codifies those patterns into guidelines that downstream content skills consume. User review gate sits between the two phases.
When to run
- User says "tone of voice", "brand voice", "voice guidelines", "writing style guide", or asks to extract voice from a URL.
- Starting a new content program for a client and the team needs one voice contract for every piece of copy.
- Existing TOV is stale (>6 months) or content has drifted from documented voice.
Don't run when: user wants visual brand identity, product messaging, to write content directly, or just a quick voice description (answer directly).
Inputs
| Input | Required | Source |
| Company URL | yes | User |
| Founder context (transcript, notes, voice preferences) | optional | User |
| Existing brand guidelines (to update) | optional | User |
| Target audience | optional | User |
Validate before proceeding: URL is accessible, site has homepage + 5-10 pages of content. If thin, flag sample-size limitation in output.
Steps
Two phases. Phase 1 outputs tov-analysis.md. User reviews. Phase 2 outputs tov-guidelines.md.
Phase 1 — Analysis
- Discover site structure — fetch homepage, extract internal links, filter same-domain only, prioritize
/about,/manifesto,/values,/blog/*,/case-studies/*,/pricing,/faq. - Scrape 15-20 pages in priority order: homepage → about/values → blog (3-5) → case studies (2-3) → pricing → FAQ → landing pages.
- Extract sentence-level patterns — average length, person (1st/3rd), question frequency, imperative usage.
- Extract paragraph-level patterns — average length, openings, transitions, evidence placement.
- Extract word-level patterns — company vocabulary, customer/industry vocabulary, banned/avoided words, modifier frequency.
- Extract structural patterns — headers (sentence vs title case), CTA placement and phrasing, proof stacking, section organization.
- Score frequency — High (80%+ of pages), Medium (40-79%), Low (<40%), Conflict (contradictory).
- Build content-type voice mapping — table of person/formality/CTA per page type (homepage, blog, case study, pricing).
- Generate voice-in-action examples — generic → on-brand transformations, drawn from actual scraped text (never invented).
- Identify inconsistencies — flag conflicts (e.g., homepage uses "I", pricing uses "we") for Phase 2 resolution.
- Document gaps — what couldn't be determined; suggest founder interview questions.
- Write
tov-analysis.mdusing the premium reference (44-section canonical scaffold; compact alternative in same file). - Present to user — Review gate. User must confirm patterns, correct misidentifications, and answer gap questions before Phase 2.
Phase 2 — Generation
- Incorporate user feedback — apply corrections, resolve inconsistencies, fill gaps with founder answers.
- Generate
tov-guidelines.mdusing the premium reference — primary reader, tone attributes with before/after, pattern library with LLM guidance, vocabulary lists, structure templates, anti-patterns. Always include an AI-speak anti-patterns section — select ≥5 relevant rows per the premium reference and embed them inline with the client's own examples, stripping row numbers and internal paths so the client's doc stands alone. 15a. Append the 'We Are / We Are Not' table per the premium reference Template 1. Source every row from Phase 1 evidence; mark confidence per row; overall confidence is the worst of any row. Minimum 5 rows, maximum 8. 15b. Append the tone-by-context matrix per Template 2. Cover the 7 default contexts (cold outbound, nurture, objection-handling, sales-call follow-up, support reply, case study, social post). Add or split rows for client-specific contexts. 15c. Append the open-questions list per Template 3. Minimum 3 questions, each with a "Why it matters" line naming the downstream unblock. - Add source attribution — every guideline traces to a source URL; frequency scores carried forward; unresolved gaps marked "TBD — requires founder input".
- Self-evaluate per the premium reference (completeness, evidence, guardrails, self-roast). Surface improvement prompts to user when checks fail.
- Save approved output as reference example if user explicitly approves — see the premium reference.
What good looks like
Evaluations
A TOV output passes when:
- 15+ pages scraped, all pattern categories (sentence, paragraph, word, structural) have findings.
- Frequency scores based on actual counts (e.g., "5 of 6 pages, 83%"), not "usually" or "often".
- Every guideline traces to a source URL.
- All examples are from actual scraped text (no invented illustrative examples).
- Inconsistencies explicitly flagged, not papered over; gaps documented with founder questions.
- No adjective-only descriptions ("friendly", "professional"); no vague guidelines ("use conversational language").
- All 6 critical questions answered: who reads / what tone sounds like / patterns repeat vs vary / words used / structure / what to refuse.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

