Public social data across platformsthrough an MCP for AI agents.
Connect once and let your agent discover and call SocQ's public social data APIs.
https://api.socq.ai/mcpAgent signal flow
Natural-language request
“Collect the latest 100 comments from this YouTube video, wait for the task, then return comment text, author, published time, and like count.”
Discover endpoint
socq_search_endpoints
socq_describe_endpointsocq_execute
POST /mcp
{ "endpoint": "…", "input": { … } }task_id
202 Accepted
{ "task_id": "…" }socq_get_task
GET /mcp
{ "task_id": "…" }results.items
200 OK
{ "items": [ … ],
"next_cursor": "…" }Hosted MCP endpoint
https://api.socq.ai/mcpTransport: Streamable HTTP
Configuration method
Choose how much of the API catalog your agent should see, select your client, and copy a configuration generated from the current SocQ platform and endpoint registry.
Keep the API key in an environment variable and connect Codex directly to SocQ's hosted Streamable HTTP endpoint.
Client
~/.codex/config.tomlConfiguration[mcp_servers.socq]url = "https://api.socq.ai/mcp"bearer_token_env_var = "SOCQ_API_KEY"tool_timeout_sec = 45What can your agent do with SocQ MCP?
Ask in natural language. Your agent selects the matching capability, reads its input requirements, submits the collection, and organizes the normalized results when the task finishes.
Backed by: youtube/commentsBacked by: x/searchBacked by: facebook-ad-library/searchBacked by: linkedin/jobsBacked by: tiktok-shop/searchBacked by: instagram/postsHow does an AI agent call SocQ MCP?
Your MCP client handles the protocol. You describe the data you need; the agent discovers or selects a tool, submits the collection, and keeps reading the same task until results are ready.
Read the MCP workflow docsAsk for the data
Describe a public social data task in natural language, including the platform, source, and desired result.
user promptDiscover or select a tool
Compact mode searches the capability registry. Platform and Exact modes can call a typed tool directly.
socq_search_endpoints → socq_describe_endpointSubmit the collection
SocQ validates the business input, creates an asynchronous collection, and returns its task identifier.
socq_execute → task_idRead and organize results
The agent continues with the same task, reads normalized result pages, and presents the answer in the requested format.
socq_get_task → results.itemsBuilt for AI agents handling cross-platform public social data
One MCP connection lets agents discover and call APIs across platforms, control which tools enter context, and continue asynchronous collection tasks through normalized results—without integrating each platform separately.
One connection, complete catalog
Compact discovery lets an agent search and execute the SocQ API catalog without configuring one MCP server per platform.
/mcp → search → describe → executeControl the tool context
Switch between discovery, selected platforms, and exact typed tools as a workflow becomes more predictable.
compact | platforms≤5 | tools≤30A real asynchronous workflow
Agents can continue the original task, read cursor-paginated normalized results, and retrieve files without submitting duplicate collections.
task_id → status → results.items → next_cursorFrequently asked questions
What is the SocQ MCP server?
It is a hosted MCP service that connects AI clients to the SocQ API catalog. Once connected, an agent can discover capabilities, submit public social data collections, and read task results.
How do I add SocQ to Codex, Claude, Cursor, or VS Code?
Codex can connect directly to the hosted Streamable HTTP endpoint. Claude Desktop, Cursor, and VS Code can run the @socq/mcp stdio bridge. Copy the matching configuration above, provide SOCQ_API_KEY, and reconnect the client.
Does one MCP URL represent only one API?
No. The default endpoint discovers and executes the full SocQ capability catalog by endpoint ID. Platform and exact tool modes can expose selected capabilities as typed tools instead.
Why doesn't the default connection expose every API as a separate tool?
Compact mode searches and describes an endpoint only when it is needed. This keeps the full catalog of schemas out of model context, reducing context usage and making tool selection more predictable.
Can I use it without an API key?
Platform, endpoint, and schema discovery are public. Collection, account, task result, and file operations require a valid SocQ API key.
Why doesn't a collection return every result immediately?
SocQ collections are asynchronous. Submission returns a task ID and can wait for up to 30 seconds. If the task is still running, the agent continues with socq_get_task instead of submitting it again.
How does MCP differ from the REST API?
Both use the same capability definitions, business inputs, billing rules, limits, and asynchronous task model. MCP adds discovery and structured tool calling, so an AI client does not need a hand-written HTTP integration.
How is MCP usage billed?
MCP does not change endpoint billing. Collections follow the same credits, API key limits, and account rules as REST. Discovery tools do not submit collection jobs.
What data can an agent access?
An agent can only call the public social data capabilities available in the SocQ API catalog. Each API page documents its supported inputs, public data scope, and normalized result fields.