Your agents. Grounded in your brand.

25 marketing agents and a growing skills catalog that read your Brand DNA over MCP, so what they make is recognizably yours from the first call.

One bearer token. Claude Code, Cursor, or any MCP client. A free 60-minute demo token below, no login.

Watch an agent use it.

One Copywriter agent, one MCP connection. It reads Paleo Pro's record through get_brand_dna_public and drafts lines only Paleo Pro would say. A recreation of a real session, shape for shape: every request and response is exactly what the wire returns.

The session below ran against the live gateway with a free demo token. Mint your own in the next section and make the same three calls.

agent session · recreation of a real call flowcaptured 2026-07-10 · token redacted

Draft the ad headline and one supporting line for Paleo Pro. Sound like the brand, not like the category.

Connecting to the jinn MCP server.

> initialize · protocolVersion 2025-06-18
{
  "protocolVersion": "2025-06-18",
  "capabilities": { "tools": {} },
  "serverInfo": { "name": "jinn", "version": "1.0.0" }
}
> tools/list
{
  "tools": [
    { "name": "ping", ... },
    { "name": "get_token_context", ... },
    { "name": "get_brand_design_md", ... },
    { "name": "get_brand_design_tokens", ... },
    { "name": "get_brand_dna_public", ... },
  ],
  "caller": { "audience": "public", "scopes": ["read"] }
}
> tools/call · get_brand_dna_public · { "slug": "paleo-pro" }
{
  "brandName": "Paleo Pro",
  "industryCategory": "CPG Protein & Nutrition Supplements",
  "archetype": "Explorer",
  "tonalAttributes": ["Direct", "Sincere", "Empowering", "Minimalist", "Adventurous", "Educational"],
  "safeWords": ["grass-fed", "pasture-raised", "whole food", "clean protein", "no fillers", ... 9 more],
  "bannedWords": ["miracle", "magic", "guilt-free", "cheat", "junk food", ... 5 more],
  "slangPolicy": "Plainspoken, approachable language. Occasional rugged/casual phrasing ('all cow, no bull'). No pop culture slang or sarcasm.",
  "positioningWedge": "Radical ingredient minimalism (four per scoop or less), uncompromising transparency, and authentic performance ...",
  "brandEnemy": "Artificial, over-processed, multi-ingredient nutritional products that disguise themselves as healthy but undermine true wellness and performance ...",
  ... 11 more fields: messagingPillars, painPoints, tribes, snapshot. Full payload 106 lines.
}

Drafting from positioningWedge · tonalAttributes · bannedWords. Avoiding "miracle", "guilt-free", "proprietary blend".

Four ingredients. Zero things you can’t pronounce.

Grass-fed, nose-to-tail nutrition in one scoop — the way your ancestors ate, minus the campfire. Read the label. That’s the whole pitch.

written from the record above
recreation · shapes and values verbatim from the live gateway · long fields elided · run it yourself below
demo token · no login60-min TTL · 3 active/IP · 5 mints/hr
$ POST /api/agents/request-demo-token
# paste into your MCP client
{
  "mcpServers": {
    "jinn": {
      "url": "https://app.jinn.works/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_JINN_MCP_TOKEN"
      }
    }
  }
}
$ tools/call get_brand_dna_public { "slug": "paleo-pro" }

Mint first. The call runs from this page with your token, against the same endpoint your client will use.

Try it now. No login.

A real token, minted here, that reads 3 showcase brands for 60 minutes. Mint, paste, and make the first call before you have told us your name.

The perimeter is honest machinery: 3 active tokens per IP, fail-closed · 5 mints an hour · a kill switch we can flip. What you get is the public projection: 5 of the 16 tools, read-only.

Same agent. Same brief. One knows the brand.

Pick a brand and sweep the divider. The generic run could belong to anyone. The grounded run was written from that brand's extracted record: its wedge, its tone, its banned words.

With Brand DNA · Copywriter

Four ingredients. Zero things you can’t pronounce.

Grass-fed, nose-to-tail nutrition in one scoop — the way your ancestors ate, minus the campfire. Read the label. That’s the whole pitch.

grounded onpositioningWedgetonalAttributesbannedWords

Four doors in.

Skills your agent installs. Personas it becomes. A gateway it connects to. An API it calls raw. Pick the door that matches how you already work.

01

Skills

MIT-licensed marketing skills your agent installs and runs locally. The catalog grows as skills ship.

Install one into your agent skills directory:

npx skills add jinn/brand-voice-content
Full instructions
02

Agents

25 marketing personas across Marketing, Brand, Content, and Design.

Connect over MCP, then put a persona on the work:

"As the Jinn Copywriter: read the brand with get_brand_dna_public, then write the launch line."
Full instructions
03

MCP

Connect Claude Code, Cursor, or any MCP client with one bearer token.

One command connects Claude Code:

claude mcp add --transport http jinn https://app.jinn.works/api/mcp --header "Authorization: Bearer $JINN_MCP_TOKEN"
Full instructions
04

API

Or POST straight to the gateway. Standard MCP over HTTP, bearer-authed.

The endpoint speaks JSON-RPC 2.0 over HTTP:

POST https://app.jinn.works/api/mcp
Authorization: Bearer YOUR_JINN_MCP_TOKEN
Full instructions

The full wire contract, tool by tool.

Read the docs
the write boundaryone write tool · fifteen read-only

> tools/call · activate_agency_pack · { "brandKey": "yourbrand.com", "packKey": "brand_strategy_audit" }

{
  "human_action_required": "checkout",
  "checkout_url": "https://app.jinn.works/agencies/checkout/..."
}

real contract · the response shape of the one write tool · a checkout URL for your human, never a charge

Agents read. Humans spend.

Every tool but one is read-only. The one that writes can queue work and hand back a checkout or intake URL, and it is architecturally unable to complete a payment on its own.

  • Entitlement is live.Your token's tier resolves against your subscription on every call, fail-closed. Paste once; upgrades and lapses take effect on the next request.
  • Every call is audited. One audit row per call, with free text redacted before hashing, so the trail can never be read back into your prompts.
  • Tokens rotate with a 24-hour overlap. The old secret keeps working while you swap the new one in. Revocation is immediate.

16 tools. Real names. Real limits.

The registry, not a roadmap. Your token's own tools/list returns exactly the set your audience and tier grant; nothing here is padded.

The 16 tools, by family

Brand DNA & access

The verified record an agent writes from, plus the tools that read the wire itself.

  • pingpublicThe smoke tool. Verifies gateway wiring end to end.
  • get_token_contextpublicWhat this token can reach: brands, scopes, tier, expiry.
  • get_brand_dnainternalThe canonical typed record: identity, positioning, voice, pillars.
  • get_brand_dna_publicpublicA bounded public projection. Competitive intel is never included.
  • get_verified_facts_statusinternalThe review-queue state: active, proposed, gaps.
  • get_marketing_truthinternalThe always-current product and strategy truth feed.

Design system

The same design contracts the Jinn products render from, read-only.

  • get_brand_design_mdpublicThe agent-readable DESIGN.md for an opted-in brand.
  • get_brand_design_tokenspublicDesign tokens as DTCG JSON: color, type, spacing, motion.
  • get_brand_kitinternalThe render-ready kit: colors, fonts, logo, name, spacing.

Agencies work

Discover, activate, and read the deliverables an owned engagement produces.

  • list_brandsinternalThe brands this token can act on, each { slug, name }.
  • list_engagementsinternalYour engagements and their per-deliverable status.
  • activate_agency_packinternal · writeThe one write tool. Queues work or hands back a human-action URL.
  • get_engagement_statusinternalPoll an engagement until each deliverable is ready.
  • get_deliverableinternalRead one rendered deliverable when it is ready.

Voice

Rewrite copy in a brand voice through the real Ghost voice-transform service.

  • list_voice_profilesinternalThe voice profiles your allowlisted brands are mapped to.
  • voice_transforminternalRewrite copy in a brand voice, with a verification report.

Public tools (5) are what a demo or self-serve token lists. Internal tools ride owner-bound and operator-issued tokens; to anything below their audience they read as unknown methods.

Token lifecycle
  • Mint. From /account/tokens. The secret is shown exactly once and stored only as a hash.
  • Rotate. A 24-hour overlap by default (48h maximum): the old secret keeps working while you swap in the new one. A revoked token cannot be rotated back to life.
  • Revoke. Immediate, from the same panel.
  • Demo. 60 minutes, read-only, showcase brands only. It expires; nothing to clean up.
Tiers
  • Demo token · Free. No login. Read-only, showcase brands, 60-minute window.
  • Connected · $99/mo. Grounds your agents on your own brand. One brand, connected over MCP.
  • Brand · $449/mo. The full brand surface for your team. Founding rate $314/mo, 30% off list and time-limited.
  • Agency · $1,499/mo. Adds the operator surface across every client brand you hold keys for. Founding rate $1,049/mo, time-limited.

Founding is the early rate: 30% off list, time-limited. Brand is $314/mo founding against $449 list; Agency is $1,049/mo founding against $1,499. No struck-through fake prices.

Audit and rate limits
  • 120 requests a minute per token. Over it, a 429 with Retry-After.
  • An audit row on every call. Free-text fields are redacted before hashing, so the stored trail cannot be dictionary-attacked back to your prompts.
  • Demo perimeter. 5 mints an hour per IP, 3 concurrently active per IP fail-closed, and a kill switch.

Want the free version first? The access checker runs one public read in seconds, no token, no account.

Try the free access checker

Three questions, straight.

Can my agent buy things?
No, by design. The one write tool queues work and hands back a human-action URL: a checkout link or an intake form. An agent can prepare a purchase; a person completes it.
What can it read?
The tools your tier and brand allowlist grant, checked live on every call. The public projection excludes competitive intel, pricing, and internal metadata.
What does it cost?
Connected is $99/mo. Brand is $449/mo, or $314/mo at the founding rate. Agency is $1,499/mo, or $1,049/mo founding. Founding is 30% off list and time-limited. The demo token is free and needs no login.

Priced by how deep the grounding goes.

Jinn Agents

Early access
  • Demo tokenFree
  • Connected$99/mo
  • Brand$449/mo
  • Agency$1,499/mo

Connected grounds your agents on your own brand. Brand adds the full team surface (founding $314/mo, time-limited). Agency adds the operator surface across client brands (founding $1,049/mo). A free demo token works against showcase brands with no login.

The rest of the suite

Priced alone
  • Ghost · content$69/mo+
  • Fama · AI-visibility auditFree

Each product priced on its own. Same brain underneath.

Early access: capabilities are feature-complete and being verified for general availability.

Runs on the Jinn Brain.

Every Jinn product starts from your Brand DNA: your voice, visuals, products, and customers, extracted from what you have actually made. Learn once. Apply everywhere.

Ground your agents.

Jinn