Docs

Check brand

Audit a product or site against brand and tone-of-voice guidelines: a Stunt Double actor reads every in-scope page like a customer and flags copy, voice, and presentation that break the rules, with evidence.

When to use

  • You have brand and tone-of-voice guidelines and want to know where the live product breaks them
  • Copy has drifted as pages shipped independently and you want a consistent voice again
  • You want brand adherence enforced continuously, not audited once

Instructions

  1. Resolve context:

    • list_workspaces() → pick the workspace
    • list_projects(workspace_id) → match the URL to a project, or let create_checklist find/create it
  2. Set up a brand-guardian actor and record the rules:

    • list_actors(workspace_id), else create_actor(...) described as a meticulous brand and copy reviewer
    • list_workspace_guidelines(workspace_id): the standard may already be written down; attach it with set_project_guideline(project_id, guideline_id) rather than typing a second copy
    • Otherwise add_project_guideline(project_id, category: "tone_of_voice" | "brand", content, title), one rule per guideline so a finding can name the rule it breaks. Guidelines are workspace-owned and reach every checklist run, design review, interview and triage for the project
    • Use add_actor_knowledge(actor_id, ...) for what only this reviewer needs to remember, not for the standard itself
    • If no guidelines were supplied, ask for them, or draft a short rule set from the strongest existing pages and confirm it before auditing
  3. Create a checklist per surface in scope:

    • create_checklist(...) for each page or flow (default scope: homepage, one core product flow, one high-traffic marketing page)
    • Phrase each check as an observable assertion a reader could verify on the rendered page:
      • "Headlines use sentence case"
      • "Copy addresses the reader as you, never the user"
      • "Error messages state what happened and offer a next step"
      • "No unexplained jargon on pricing"
    • 4–8 checks per checklist
  4. Run and collect evidence:

    • run_checklist(checklist_id) → poll get_checklist_run(run_id) until terminal; read the evidence per check
  5. Report deviations grouped by page:

    • Each with the broken rule, the offending copy or element as evidence, and a suggested rewrite in the correct voice
  6. Make it standing:

    • Offer create_workflow on a schedule or deploy trigger so the brand stays on-voice as pages change (see setup-guardrails)

Example output

Brand & voice audit: acme.com

Pricing page:
  [FAIL] "The User can upgrade at any time": addresses reader in third person.
         Rewrite: "You can upgrade any time."
  [FAIL] Headline "GET STARTED NOW!!!": all caps + exclamation, breaks sentence-case + no-exclamation rules.
         Rewrite: "Get started"

Onboarding:
  [PASS] Sentence-case headlines throughout
  [FAIL] Error "Invalid input": states no cause and no next step.
         Rewrite: "That email is already in use. Try signing in instead."

3 deviations across 2 pages. Offer: schedule this checklist weekly to hold the voice.

Tips

  • Store the rules on the actor, not just in the checklist. Actor knowledge makes every future run enforce the same standard automatically.
  • Checks must be reader-observable. Actors see the rendered page, so phrase rules by what a customer would read, not by CSS or component names.
  • One checklist per surface keeps failures attributable to a page.

Edit on GitHub

On this page