@aryaminus/socials-mcp
Enables social media creators to pull TikTok, Instagram, Facebook, and YouTube analytics into a persistent vault, receive weekly digests, review scripts, build media kits, and draft brand outreach using official APIs only.
README
Socials Assistant
MCP server + agent skills for social media creators. Pull your own TikTok, Instagram, Facebook, and YouTube analytics into a permanent vault, get weekly digests, review scripts, build media kits, and draft brand outreach with real numbers — from any AI agent.
Official APIs only. No scraping. No ToS risk.
Quick start — pick one
1. One prompt (easiest)
Copy the prompt from docs/SETUP-PROMPT.md and paste it into Claude, ChatGPT, Gemini, Codex, Cursor, or any MCP-compatible agent. It deploys the server, connects it to your agent, installs the 6 skills, walks you through platform OAuth, and runs your first digest. You click consent screens; the agent does the rest.
2. Cloud, zero install
Then open https://<your-worker>.workers.dev/setup for the credential checklist. Full journey: docs/GET-STARTED-CLOUD.md.
3. npx
npx @aryaminus/socials-mcp onboard
Installs nothing globally — prints guided setup for platform credentials, then serves MCP on stdio/HTTP.
4. Local from source
git clone https://github.com/aryaminus/socials-assistant && cd socials-assistant
./setup.sh && node apps/mcp/bin/socials-mcp.js onboard
What you get
| 20 MCP tools | connect · snapshot · vault query · period compare · top content · audience · digest · media kit · pipeline · outreach log |
| 6 skills | socials-connect · weekly-digest · script-review · publish-package · brand-outreach · media-kit |
| History vault | platforms expire analytics; your SQLite (local) or D1 (cloud) vault doesn't |
| TikTok CSV import | retention, traffic sources, search terms — Studio-only data via one command |
| Draft-first outreach | real numbers from your vault; nothing sends without your approval |
Connect your agent
| Agent | How |
|---|---|
| Claude.ai / Desktop | Settings → Connectors → custom connector → paste MCP URL |
| Claude Code | claude mcp add --transport http socials <URL> |
| Codex | [mcp_servers.socials] url = "<URL>" + codex mcp login socials |
| Cursor | Settings → MCP Servers → paste URL |
| Gemini CLI | mcpServers block in ~/.gemini/settings.json |
| Everything else | agents/README.md |
Docs
| Doc | Read when |
|---|---|
| docs/SETUP-PROMPT.md | you want the one-prompt path |
| docs/GET-STARTED-CLOUD.md | deploying to Cloudflare |
| docs/onboarding-google.md / meta / tiktok | creating platform apps for OAuth |
| docs/CAPABILITIES.md | checking what data each platform exposes |
| docs/hosting-cloudflare.md | manual cloud deployment details |
| docs/automation.md | scheduling weekly snapshots + digests |
| AGENTS.md | env vars, CLI commands, release process |
| CONTRIBUTING.md | developing on this repo |
Security
Tokens encrypted at rest (AES-256-GCM) · official APIs only · vault local by default · outreach never auto-sends. Details: SECURITY.md.
MIT © 2026 aryaminus
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。