Layers Marketing MCP Server已验证已精选

Give your agent marketing superpowers

Layers MCP is the endpoint your coding agent calls to run growth for your product. It onboards the product, keeps the map of accounts you compete with for attention, finds what to post, plans the test, renders the video or the slideshow, delivers it, reads the result, runs paid campaigns and Layers-operated accounts, and stops everything on one call.

MCP endpoint: mcp.layers.com/mcp · Site: https://growth.layers.com · Docs: https://layers.com/docs/mcp

Paste this to your coding agent

Set up Layers Growth for me so you can plan, render and deliver content for this product from here.
1. Install the CLI: run `npm install -g @layers/cli`.
2. Check that the README opens with a plain sentence saying what this product is, and write one if it does not. Setup sends that single line as the product brief, and it stops with exit code 1 when it cannot find one.
3. Set it up: run `layers setup` from this repository. It opens my browser once for sign-in and waits for me; it writes the `.mcp.json` entry and the Layers skill into the repo and stores the session in my OS credential store, so nothing secret lands in the repo. It may ask for the product's public URL, which is its website, its App Store page, or its Google Play page; pass that question to me.
4. Read the skill at `.agents/skills/layers/SKILL.md`, then tell me what to post first.
Let me know when it is ready.

Your agent runs the commands. The browser sign-in is the only step that needs a person.

What your agent can do

Twenty-seven named outcomes. One call does one job and returns one envelope with the cost, the freshness of the data behind it and the next tools worth calling. All twenty-seven reach your agent the moment layers setup binds a project. There is no second install, no per-feature switch and no waiting list.

Growth — onboard_product (free) · manage_growth_landscape (free; 6 credits per forced refresh or rebuild) · find_growth_opportunities (free) · plan_content_experiment (6 credits per new plan; creative edits free) · generate_influencer (free) · render_content (25 credits per image, 85 per slideshow, 200 per 10-second video, 400 per 20-second video; unusable output refunded) · deliver_content_experiment (free to prepare, 5 credits per delivery target) · track_growth_experiment (free) · review_growth_results (free; 1 credit per forced provider read) · halt (free) · cancel_trial_auto_renewal (free) · send_feedback (free) · add_brand_asset (free) · install_growth_measurement (free) · generate_tracking_link (free)

Research · read the platform you are growing on, one credit a call — research_tiktok_creator · research_tiktok_creator_posts · research_tiktok_search · research_tiktok_hashtag · research_tiktok_trending_sounds

Managed distribution · Layers-operated TikTok accounts, billed in dollars — plan_managed_distribution (free) · provision_approved_managed_accounts ($150 per account per month from the cash wallet) · release_managed_accounts (free)

Paid media · TikTok is the built provider — plan_paid_campaign (free) · create_approved_campaign_draft (10 credits per campaign) · launch_approved_campaign (free; ad spend billed by the platform) · apply_approved_campaign_change (free, or 5 credits per campaign changed)

Fourteen resource families under layers://, all free to read: account, account/credit-events, workspace, landscape, opportunities, experiments, artifacts/, influencers, music, brand-assets, jobs, skills/v1, campaigns, managed-accounts, measurement.

Layers never receives your source code.

How people use it

"What should I post tomorrow for my meditation app?" — find_growth_opportunities returns the standing ranked queue, each test carrying the post it came from and how that post beat its own account's normal.

"Which of our competitors' posts broke out this week?" — manage_growth_landscape keeps the tracked set and layers://landscape returns it ranked, with the source evidence and when it was observed.

"Plan a three-hook test for our onboarding video and send the winner to TikTok drafts." — plan_content_experiment writes the plan with one controlled variable, render_content makes the variants, and deliver_content_experiment puts the winner in the draft inbox.

"Track the video I posted from my own account and tell me if it beat our normal." — track_growth_experiment registers the post with its hypothesis, and review_growth_results scores it against the account's own top quartile.

"Review last week's experiment and tell me what to double down on." — review_growth_results reads the evidence, records the verdict with its uncertainty, and names the next controlled test to run.

"Stop everything on this product now." — halt stops queued and running work in the scope you named and returns a receipt for every target it touched.

"Make me a face for the app and put her in a reaction video." — generate_influencer creates the persona for free and hands back a slate page of every face in the project, then render_content casts the completed one in a reaction_demo at two hundred credits.

"Set Layers up for this repo and connect our TikTok account." — onboard_product derives the brief from the repository, the website and the store listing, then returns the connection page. It sends the brief fields and a digest. No file, path or diff leaves your machine.

Credits

One balance per organization, and the same operation costs the same on every plan. Pay as you go is $0 a month with no included credits, so the cheapest way in is to buy credits and spend them. A trial starts with 100 credits.

Packs, at two cents a credit: 500 for $10, 1,000 for $20, 2,000 for $40, 4,000 for $80. A paid plan buys the same four sizes cheaper, at 500 for $5, 1,000 for $9, 2,000 for $17 and 4,000 for $29. Packs add to the same balance and do not expire.

Every charged tool takes quoteOnly, which returns the exact maximum and runs no provider call and no model call. A render that produces nothing usable is refunded. Auto-reload is managed at https://app.layers.com. Ad spend and managed accounts are separate money and never consume credits.

Manual configuration

The two commands, run yourself, from your repository root:

npm install -g @layers/cli
layers setup

The stdio server, which is the recommended path. Your MCP client spawns layers mcp, which is the only reader of the session in your credential store:

{
  "mcpServers": {
    "layers": { "command": "layers", "args": ["mcp"] }
  }
}

The remote endpoint. Streamable HTTP at https://mcp.layers.com/mcp. POST one JSON-RPC message per request, and send the session layers login issues as the bearer:

POST https://mcp.layers.com/mcp
Authorization: Bearer <layers session>
Content-Type: application/json
Accept: application/json, text/event-stream

Links

• Documentation: https://layers.com/docs/mcp
• Open Layers: https://app.layers.com
• llms.txt: https://growth.layers.com/llms.txt · llms-full.txt: https://growth.layers.com/llms-full.txt
• Repository: https://github.com/layers/mcp