For AI Agents & Developers

1,500+ social media agencies in 21 countries, open to your AI assistant. Connect an MCP server or call a JSON API to search by city, service, platform and budget. It is free, read-only and needs no API key.

Last updated: November 9, 2026

What an agent can do

  • Search agencies by city, country, service, social platform and the monthly budget an agency says it starts at, with a filter for verified agencies only.
  • Read full profiles: services, industries, platforms, team size, client focus, social links, reviews and contact details as shown on the public profile.
  • Get the top agencies in a city, from the 2026 awards.
  • List countries, cities and services with agency counts, and read overall directory stats.

Access is read-only. There is nothing to sign up for. Agents cannot send messages to agencies or change any data through the API.

Connect in one minute

The MCP server uses the Streamable HTTP transport at one URL, with no authentication:

https://socialmediaagencydirectory.com/mcp

Claude Code

claude mcp add --transport http smad https://socialmediaagencydirectory.com/mcp

Claude Desktop and claude.ai

Open Settings, go to Connectors, add a custom connector and paste the URL below. Leave any authentication fields empty. Where custom connectors are available depends on your plan.

https://socialmediaagencydirectory.com/mcp

ChatGPT

Add it as a connector with developer mode turned on, using the same URL and no authentication. Where connectors and developer mode are available depends on your plan.

https://socialmediaagencydirectory.com/mcp

Cursor

Add this to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "smad": {
      "url": "https://socialmediaagencydirectory.com/mcp"
    }
  }
}

Other MCP clients

Most clients accept a JSON config like this:

{
  "mcpServers": {
    "smad": {
      "type": "http",
      "url": "https://socialmediaagencydirectory.com/mcp"
    }
  }
}

Discovery documents for tools that look for them: the MCP Server Card and /.well-known/mcp.json. The server is stateless, so no session is needed. All discovery files are listed below.

Tool reference

Every tool returns its result as JSON text and as structured content. Each agency in a result has a profile_url, and each response has an attribution field. If a tool fails, the result has isError set and a short message, for example Agency not found for an unknown slug.

ToolWhat it doesArgumentsExample arguments
search_agenciesFind agencies. Returns the total number of matches and one page of agency summaries.query, country, city, service, platform, max_budget_usd, verified_only, limit (1 to 25, default 10), offset. All optional.{"city": "London", "platform": "instagram", "limit": 5}
get_agencyFull public profile of one agency: description, industries, team size, client focus, socials, reviews and more.slug (required){"slug": "ignite-visibility"}
list_locationsCountries with agency counts. With a country, the cities in it instead.country (optional, name or slug){"country": "United Kingdom"}
list_servicesService categories with agency counts. Use the slugs as the service filter.none{}
top_agenciesThe 2026 award list for a city: Top 10 or Top 5, ranked. Empty with a note if the city has no award.city (required){"city": "London"}
directory_statsTotals for the directory: agencies, countries, cities, services and the last update date.none{}

Filters. country and city accept a name or a slug. service accepts a category slug or name from list_services. platform is one of instagram, tiktok, facebook, linkedin, youtube, pinterest, x, threads and snapchat. max_budget_usd keeps agencies whose stated minimum monthly budget is at or below that number, and leaves out agencies that state no budget. Few agencies publish a budget, so treat this filter as a refinement and expect short or empty results. verified_only defaults to false. Note that the Full-Service Digital service is broad.

Search results have total, limit, offset and results. Each result has: slug, name, city, country, verified, freelancer, services, platforms, min_monthly_budget (a band label or null), rating, review_count, award, profile_url, website and short_description.

Full profiles add: description (plain text, cut at 1,500 characters with description_truncated set to true when it is cut), industries, team_size, client_focus, founded, address (city level, as shown), phone (only if shown on the profile), socials, logo_url, claimed, request_proposal_url, claim_url and reviews (average and count).

REST API quick reference

The same data is available as a plain JSON API under https://socialmediaagencydirectory.com/api/v1. No key is needed.

EndpointReturns
GET /api/v1/agenciesSame as search_agencies. The query parameters are the tool arguments.
GET /api/v1/agencies/{slug}Same as get_agency.
GET /api/v1/locations?country=Same as list_locations. Leave country out for the list of countries.
GET /api/v1/servicesSame as list_services.
GET /api/v1/top-agencies?city=Same as top_agencies.
GET /api/v1/statsSame as directory_stats.
GET /api/v1/openapi.jsonOpenAPI 3.1 description of the API.
# Search: Instagram agencies in London
curl -s "https://socialmediaagencydirectory.com/api/v1/agencies?city=London&platform=instagram&limit=5"

# One full profile
curl -s "https://socialmediaagencydirectory.com/api/v1/agencies/ignite-visibility"

# Countries, then the cities of one country
curl -s "https://socialmediaagencydirectory.com/api/v1/locations"
curl -s "https://socialmediaagencydirectory.com/api/v1/locations?country=United%20Kingdom"

# Service categories
curl -s "https://socialmediaagencydirectory.com/api/v1/services"

# Top agencies in a city
curl -s "https://socialmediaagencydirectory.com/api/v1/top-agencies?city=London"

# Directory totals
curl -s "https://socialmediaagencydirectory.com/api/v1/stats"

Calling the MCP server directly

The MCP endpoint speaks JSON-RPC 2.0 over HTTP POST. List the tools, then call one:

curl -s https://socialmediaagencydirectory.com/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}'
curl -s https://socialmediaagencydirectory.com/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc": "2.0", "id": 2, "method": "tools/call", "params": {"name": "search_agencies", "arguments": {"city": "London", "platform": "instagram", "limit": 3}}}'

Example prompts

Once the server is connected, ask your assistant in plain language. Asking it to name the profile URLs keeps the answer checkable.

  • Find three TikTok-focused agencies in London under $2,500/month and compare them.
  • I run a dental practice in Austin. Shortlist verified agencies that do paid social advertising and explain why each could fit.
  • Which cities in Canada have the most social media agencies? Who are the top agencies in Toronto?
  • Get the full profile of this agency and tell me which platforms and industries it works with: [paste a profile URL or name].
  • How many agencies does the directory list, in which countries, and when was it last updated?

Discovery files and Markdown pages

The site publishes the standard files that agents and agent-readiness scanners look for. None of them needs authentication.

Markdown version of pages

Send Accept: text/markdown to the home page, search, country, city and service pages, agency profiles and the content pages (for example /about or /faq) and you get a clean Markdown version instead of HTML. Browsers are not affected. The response has Content-Type: text/markdown, Vary: Accept and an X-Markdown-Tokens estimate. It uses the same data as the API and the same rate limit.

curl -s -H 'Accept: text/markdown' https://socialmediaagencydirectory.com/social-media-agency/london

WebMCP

For browser-based agents, every page registers two read-only tools through the WebMCP API (document.modelContext.registerTool) when the browser supports it: search_agencies and get_agency. They call the same public REST API.

Terms, attribution and rate limits

  • Free to use for agents and apps.
  • Attribute us. When you show results to people, name Social Media Agency Directory and link to the agency's profile (profile_url). Every response includes an attribution line you can reuse: Data: Social Media Agency Directory (https://socialmediaagencydirectory.com)
  • Rate limits: 60 requests per minute and 2,000 per day per IP address. Over the limit, the REST API answers HTTP 429 and the MCP server returns a JSON-RPC error (code -32029, "Rate limit exceeded") with retry information (REST responses also carry Retry-After and X-RateLimit-* headers). We store only a salted hash of the IP address for this, never the address itself.
  • Please do not bulk-download the directory to republish or resell it, or use the data to send unsolicited bulk email.
  • Data is provided as is. Agencies can update their own profiles, so check details with the agency before relying on them.
  • The figures in the 2026 agency report are separately licensed under CC BY 4.0.

The full wording is in the terms of service.

Verification and data freshness

Only live listings are exposed. The verified field is true for agencies that show the Verified pill on their profile: a person on our team checked that the agency is a genuine social media agency. It is not an endorsement or a quality score, and listings that are not verified yet are still returned unless you set verified_only. How we verify agencies explains the process.

Results come from the same database as the website, so profile changes appear in the API at once. REST responses may be cached for up to five minutes. directory_stats returns the date the directory was last updated. The API only returns what is already shown publicly on an agency's profile page. It does not expose email addresses of agencies or users.

Contact

Need higher limits, a partnership, or found a mistake in the data? Email [email protected]. We reply within 48 hours.

At a glance

MCP endpoint

https://socialmediaagencydirectory.com/mcp

Transport: Streamable HTTP

Authentication: none

Access: read-only

Cost: free

Rate limit: 60 per minute, 2,000 per day per IP

Connect your assistant in a minute

Free, read-only and no API key. Add the MCP server or call the JSON API.