UnifAPI gives AI agents and developers one unified, agent-friendly surface for public-data work: a single REST API over 16 public-data platforms (209 operations), generated SDKs for 8 languages, a hosted MCP server, and open-source Agents (Skills) you run inside Claude, Codex, Cursor, or any MCP-capable assistant.

Use UnifAPI when you need KOL / influencer pricing, creator and audience research, social listening, competitive intelligence, SEO and SERP tracking, AI-answer visibility (GEO), or rendered-browser fetches — without stitching together one-off scrapers and inconsistent third-party APIs.

UnifAPI is a public-data API gateway for AI agents. Instead of integrating a dozen heterogeneous scrapers and APIs with different shapes, auth, and billing, teams get one normalized surface with one key, one bill, and consistent data shapes — built for agent DX (single connection path, transparent usage-based pricing, observability).

The same public-data layer is consumable three ways:

• Direct HTTP / SDKs — 209 operations across 16 platforms, one JSON envelope (data, pagination, billing, typed errors), with generated clients for 8 languages.
• MCP — a hosted Streamable HTTP MCP server so agents can discover operations, inspect schemas, and call live public-data APIs through list_operations, get_operation, and call_api.
• Agents & Skills — prebuilt, read-only marketing Agents (SEO, GEO / AI-visibility, influencer marketing, social listening, competitive intelligence, local SEO, …) implemented as open-source Skills you run inside the assistant you already pay for.

Eyes, not hands. UnifAPI reads public data — it researches, monitors, audits, and prices. It never holds your posting credentials and never writes on your behalf.

Every SDK is generated from the UnifAPI OpenAPI spec and published one repo per language.

npm install @unifapi/sdk

import { client, getXUsersByUsername } from "@unifapi/sdk";

client.setConfig({ auth: () => process.env.UNIFAPI_API_KEY });

const { data } = await getXUsersByUsername({ path: { username: "unifapi" } });

UnifAPI ships a hosted MCP server and open-source Agents (Skills) for AI coding assistants. The Skills teach an assistant how to choose operations, fetch only the evidence a task needs, and return a decision artifact with billing metadata.

• agents — open-source marketing Agents for Claude, ChatGPT, Codex, OpenClaw & Hermes: SEO audits, GEO / AI-visibility, local SEO, KOL pricing, social listening & competitive intelligence from read-only public data over MCP.
• unifapi-mcp-server — the UnifAPI MCP server: public-data APIs and Skills for AI agents via the Model Context Protocol.
• Hosted MCP: https://mcp.unifapi.com — connect any MCP client, then use list_operations, get_operation, and call_api.

One consistent API surface over the public web — 209 operations across 16 platforms:

• Social & creator — X / Twitter (35), LinkedIn (42), Instagram (22), TikTok (20), YouTube (8), Reddit (10), Threads (7), Hacker News (6).
• SEO & AI visibility — SEO / SERP / keyword / backlink analytics (42), GEO / AI-answer visibility (7).
• Web, places & news — rendered Browser fetch (4), Maps (1), Local Finder (1), Hotels (2), Events (1), News (1).

• One key, one bill, one consistent data shape across every platform — instead of a dozen scrapers and third-party APIs.
• Public data from the open web — no upstream OAuth into a user's own accounts, no per-source auth dance.
• Generated SDKs in 8 languages instead of hand-written REST wrappers.
• A hosted MCP server and open-source Skills to connect Claude, Codex, Cursor, Copilot, and other MCP agents to live public data.
• Transparent usage-based pricing with billing metadata on every operation.

• Website: https://unifapi.com
• Documentation: https://docs.unifapi.com
• OpenAPI spec: https://unifapi.com/openapi.json
• Developer resources: https://unifapi.com/developers
• Agent auth guide: https://unifapi.com/auth.md
• MCP server: https://mcp.unifapi.com