Access live public data from 25 social platforms through one MCP interface and a unified response schema.
Copy the AI prompt to install this server into Claude Code, Cursor, or another agent β or use 1-click editor setup below.
We haven't yet run this listing's install command through our automated sandbox check. This isn't a red flag β we're steadily working through the catalog.
π‘ Paste the JSON block into your client's configuration file under mcpServers, then restart the application.
Inspect callable tools, capabilities, and parameters exposed to AI agents by InsightSocial.
The InsightSocial MCP server exposes the InsightSocial API to AI agents. The API covers 25 platforms, including Instagram, TikTok, LinkedIn, Facebook, YouTube, X, Reddit, Threads, Pinterest, Bluesky, Truth Social, Snapchat, Telegram, Twitch, Kick, Rumble, Kwai, Douyin, Xiaohongshu, Quora, TikTok Shop, Weibo, Zhihu, Substack, and Nextdoor.
Its catalogue contains more than 300 endpoints. Depending on the endpoint, an agent can retrieve profiles, public email data, posts, reels, comments, followers, likers, hashtags, search results, company pages, jobs, advertisements, transcripts, shop products, and local business listings. Endpoint availability and inputs vary by platform, so agents can search the catalogue and inspect an endpoint before making a paid request.
The InsightSocial MCP server offers five MCP tools: search_endpoints, describe_endpoint, call_endpoint, read_result, and get_credits. Search and description operations are free. call_endpoint makes the API request, stores the complete response, and returns a result identifier with a trimmed view and a command for fetching another page. Each additional page is a separate call.
Saved responses can be reshaped locally with jq expressions, selected fields, item limits, or a structural outline. These operations do not make another network request or incur another charge. The API requests schema 2 for every call, giving supported platforms a shared field layout while recording fields that could not be populated in an unavailable list. A dry-run parameter can return a price quote without charging the account.
Run npx -y insightsocial login to authenticate through a browser, or provide an existing key with --api-key. The key is saved locally and the MCP process reads it without placing the secret in the MCP configuration. INSIGHTSOCIAL_API_KEY can also be supplied in the server environment.
Start the local server with npx -y insightsocial mcp. The standard MCP configuration uses npx, the -y option, and the insightsocial mcp arguments. npx -y insightsocial init --yes can write setup entries for Claude Code, Codex, and Cursor. The hosted endpoint is available at https://api.insightsocial.app/mcp; clients that support its authentication flow can use OAuth, while other clients can send a bearer key.
INSIGHTSOCIAL_BASE_URL changes the API base URL from its default. INSIGHTSOCIAL_HOME changes the default local directory for configuration, catalogue cache, and MCP results. Without a key, endpoint search and description remain available, but data calls require authentication.
search_endpoints: locate endpoints by keywords and platform.describe_endpoint: show inputs, accepted values, an example, and the endpoint price.call_endpoint: fetch and save complete API responses while returning a reduced view.read_result: inspect saved data with jq, field selection, item limits, or size summaries.get_credits: check account balance, plan, and usage.The command-line interface also supports listing endpoints, running paths directly, viewing the latest saved result, checking credits, paging through responses, and supplying idempotency or freshness options.
Requests are priced per endpoint call. Empty results, failed calls, and idempotent replays are free, but each requested page is charged separately. The shared schema does not guarantee that every platform supplies every field; unavailable values are marked as unknown rather than treated as zero. The service provides public data, and the available information depends on each endpoint and platform.
The local server uses stdio. Results are saved in full locally even when the returned MCP view is trimmed, so agents can reduce context size without discarding the original response.
No reviews yet β be the first to share how this listing worked for you.
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
[](https://allmcps.com/mcp/insightsocial)<a href="https://allmcps.com/mcp/insightsocial"><img src="https://allmcps.com/api/badge/insightsocial?style=directory" alt="InsightSocial on AllMCPs" /></a>