goodnight000/api-anything

Turn any website into an API your agent can call. Learns from the site's own frontend, heals itself when the site changes.

What it solves

API Anything allows AI agents to interact with websites as if they were APIs, eliminating the need for agents to manually drive a browser (like using Playwright or Puppeteer). This reduces the time, token usage, and cost associated with browser-based automation while improving reliability and speed.

How it works

  • Learning Phase: The tool uses headless Chrome to record network requests made by a website when provided with two example inputs. It identifies the parts of the request that change based on the input and creates a template with slots for those parameters.
  • Execution: When an agent calls an operation, the tool fills the slots and sends the request using Node's fetch. If plain HTTP is blocked, it falls back to fetching from within a Chrome page or running the page itself to read the response.
  • Session Management: It imports cookies and tokens from existing browser profiles (Chrome, Arc, Brave, Edge, Chromium, Firefox) to handle authentication without requiring passwords to be typed manually.
  • Repair: If a request fails due to rotated IDs or moved fields, the tool can re-learn the operation to update the template.

Who it’s for

Developers and AI agent creators who want their agents to perform web-based tasks (like searching flights or books) more efficiently and cheaply than traditional browser automation.

Highlights

  • Performance Gains: Benchmarks show agents using API Anything answer tasks 1.5x to 2x faster and use roughly half the tokens compared to browser-driving agents.
  • MCP Support: Integrates with the Model Context Protocol (MCP) for easy connection to agents like Claude Code or Codex.
  • Automated Mapping: Includes a "skill" that allows an agent to autonomously scout a site and build new operations based on a user's intent.
  • Pre-built Operations: Comes with 22 read operations for popular sites including Google Flights, Hacker News, X, Instagram, YouTube, Airbnb, Amazon, and LinkedIn.

Related

  • Project
    undertheseanlp/underthesea

    An open-source agentic AI toolkit and Vietnamese NLP library that provides multi-provider LLM support and specialized text processing tools for the Vietnamese language.

  • Project
    haydenbleasel/blume

    A zero-config documentation framework that turns Markdown/MDX folders into AI-ready documentation sites with built-in support for MCP servers and agentic tooling.

  • Project
    a2aproject/a2a-samples

    A collection of code samples and demonstrations for the Agent2Agent (A2A) protocol, designed to enable interoperability and communication between different AI agents.

  • Project
    qusong0627/QuantMind

    QuantMind is an open‑source AI‑first quantitative‑trading platform that combines Microsoft Qlib, RD‑Agent LLM factor mining, and a chat‑bot (QuantBot) reachable from WeChat, QQ, Feishu, etc. It offers end‑to‑end data → factor → model → back‑test → simulated trading pipelines, 13 built‑in model types, autonomous factor evolution, and Docker‑based deployment.