Skip to content
Streamline Social

Streamline Social

For developers

Connect an AI assistant to a brand's Streamline Social analytics, read only.

What this connects

Streamline Social exposes one MCP (Model Context Protocol) server and a matching REST API, both read-only, so an AI assistant like Muse or Claude can answer questions about a brand's content performance, audience, peers, and comments using the same data this app shows a signed-in team. Nothing it returns can be used to post, message, or change anything - every tool and every endpoint only ever reads.

MCP endpoint: https://streamlinesocial.ai/api/mcp

Authentication

Two ways to connect, depending on the client:

  • Sign in with Streamline Social (OAuth). For a connector Meta or Anthropic lists in a directory, like Muse or Claude - the assistant sends the person through a normal sign-in and a consent screen ("allow this app to read your brands' analytics?"), and nothing is ever pasted by hand. This is standard OAuth 2.1 with PKCE and dynamic client registration; most MCP clients handle it automatically once they see this server's metadata.
  • An API key. For a custom connector - one you're wiring up yourself, not from a directory. Create a key in Settings > API keys (an owner only) and pass it as Authorization: Bearer ss_live_.... A key is scoped to exactly one brand for as long as it's live.

Tools

Every tool takes an optional brand argument (a slug or name) to pick which brand to read - an API key ignores it unless it names that key's own brand; an OAuth sign-in with several brands defaults to the earliest one.

list_brands

Every brand this connection can read: name, slug, self-account handle and platform, and role.

get_brief

The latest weekly performance brief, plus the structured numbers behind it.

find_proof

Search past posts, peers, hashtags, and discovered accounts for proof a content idea has worked.

whats_asked

Recurring comment themes, with verbatim quotes and content ideas.

audience_overview

The audience read, accounts worth studying, and the engagement playbook.

peers

The tracked peer table: followers, cadence, engagement, and format mix.

what_to_post

A direct recommendation: patterns to do more of and less of, plus open audience questions.

REST endpoints

The same data, one call per topic, under https://streamlinesocial.ai/api/v1:

GET /api/v1/brief

The latest weekly brief and its chart data.

GET /api/v1/proof?topic=...&format=reel|carousel|image|any

Proof a content idea has worked before.

GET /api/v1/questions

Recurring comment themes.

GET /api/v1/audience

The audience read, recommendations, and playbook.

GET /api/v1/peers

The peer table.

GET /api/v1/what-to-post

Do-more/do-less patterns and open questions.

Every REST endpoint also takes an optional ?brand= query parameter, same rule as the tools above.

Rate limits

60 requests per minute per credential (an API key, or a signed-in person), enforced per server instance - good enough for the "an agent asks a handful of questions" traffic this API is built for. Going over it returns 429 with a Retry-After header.

Example prompts

  • "What should I post this week?"
  • "Has a carousel about homesteading worked for me before?"
  • "What are people asking in my comments?"
  • "Which accounts should I study for my faith and family pillar?"
  • "How did my audience respond last week compared to the week before?"
  • "Which of my brands is growing fastest?"

Connect in Muse

  1. 1In Muse, ask it to add a custom connector (or open its connector settings directly).
  2. 2Give it the MCP URL: https://streamlinesocial.ai/api/mcp
  3. 3Sign in with your account when prompted, and allow the connection on the consent screen - or paste an API key instead if you created one in Settings.
  4. 4Ask it one of the example prompts above to confirm it can read your data.

Connect in Claude

  1. 1Go to Claude's Settings > Connectors > Add custom connector.
  2. 2Enter the MCP URL: https://streamlinesocial.ai/api/mcp
  3. 3Sign in and allow the connection when prompted, or paste an API key.
  4. 4Ask Claude one of the example prompts above.

For MCP client implementers

This server follows the MCP Authorization spec (2025-11-25). Protected resource metadata (RFC 9728) is at /.well-known/oauth-protected-resource/api/mcp and at the origin root; a request without a valid credential gets a 401 with a WWW-Authenticate header pointing at that metadata. The authorization server is Supabase Auth's own OAuth 2.1 server, which supports PKCE (S256) and dynamic client registration - most standard MCP clients need no manual configuration beyond the MCP URL above.

{
  "resource": "https://streamlinesocial.ai/api/mcp",
  "authorization_servers": ["https://htkhoxggqjrksxojskdh.supabase.co/auth/v1"],
  "bearer_methods_supported": ["header"]
}