Twenty minutes per repo hand-writing CLAUDE.md
New services, forks, and experiments each need the same voice and token rules copied by hand. Small drift compounds into inconsistent agents and UI.
For Developers
Generate CLAUDE.md and .cursorrules from structured mythos, not from fragile copy-paste. Keep repos, IDEs, and models aligned when brand evolves.
See pricing · All export formats · Integration guides (Claude, Cursor)
Text files in repo roots are the right interface for agents, but they should be generated from a single governed source, not retyped under deadline pressure.
New services, forks, and experiments each need the same voice and token rules copied by hand. Small drift compounds into inconsistent agents and UI.
You fix CLAUDE.md in one repo and forget another. Six months later, three versions of ‘truth’ ship to production with no audit trail.
Brand is infrastructure, but it lives in PDFs and Slack threads. You cannot tag, diff, or roll back the instructions your models actually read.
New engineers do not know the rules exist. Cursor suggests off-brand code paths; Claude drafts tone-deaf copy because nothing loads automatically.
BrandMythos keeps human-facing guidelines and machine-consumable rules in sync so your stack stays honest as brand evolves.
Paste a site URL or ingest PDFs and Figma. BrandMythos extracts structure and emits repo-ready markdown your agents load on day one.
One mythos in BrandMythos; push CLAUDE.md and .cursorrules to GitHub, GitLab, or Bitbucket when your source of truth changes.
Treat exports as artifacts: tag releases, review diffs in PRs, and roll back when brand updates misfire.
Same structured data powers .cursorrules, system prompts, and hosted guides, with no parallel maintenance per tool.
Shared workspaces with visibility into who changed tone rules, tokens, or safety rails before they hit production repos.
Fetch mythos JSON, trigger exports, and wire CI so brand updates propagate without a human copy-paste step.
URL crawl, PDF, Figma, or Drive. BrandMythos normalizes voice, visual, and messaging into structured mythos.
Emit CLAUDE.md, .cursorrules, tokens, prompts. Push via Git integration or download for review.
Cursor reads .cursorrules; Claude reads CLAUDE.md; ChatGPT uses exported system prompts, all from the same versioned snapshot.
Illustrative snippets; your exports reflect your actual mythos and repository conventions.
CLAUDE.md (excerpt)
# Acme: agent brief## VoiceDirect, evidence-first. No ALL CAPS emphasis.## VisualPrimary #9C4221 for hero CTAs only. Body: DM Sans.## RulesDO: short paragraphs, numbered steps.DON'T: hype adjectives without cited proof..cursorrules (excerpt)
# Cursor: Acmevoice: see CLAUDE.md in repo rootcomponents: PascalCase under src/components/colors: var(--color-*) onlyspacing: --space-2 | --space-4 | --space-8a11y: landmarks; visible focus on controlsWebhook-driven sync (illustrative)
api/mythos-webhook.ts
// CI: POST mythos.version.tagged (excerpt handler)app.post("/hooks/mythos", async (req, res) => { const { mythosVersion, exports } = req.body; await queueRepoSync({ targets: exports.map((e) => e.repoPath), artifacts: ["CLAUDE.md", ".cursorrules", "AGENTS.md"], tag: mythosVersion, }); return res.sendStatus(204);});Connect version control once. When mythos changes, regenerate and open PRs so reviewers see a single coherent diff, not a surprise paste in Slack.
Example sync targets
brand-mythos → services/web/CLAUDE.md
brand-mythos → services/web/.cursorrules
brand-mythos → packages/ui/CLAUDE.md
Tag each release with the same mythos version across services.
Start on Free with one scan. Move to Pro when you need multiple workspaces, unlimited scans, and API hooks, aligned with our published tiers.
Prove the pipeline
$0
Teams shipping weekly
$199
/month (or $1,908/year, save 20%)
Need Business (15 brands, white-label, MCP roadmap priority)? Compare plans or talk to us.
No credit card to try the signup flow. Your first scan shows how voice, visual, and messaging compress into files your stack can own.