SuperAPI AI Directory

Obscura

Obscura is an open-source headless browser engine for AI agents and web scraping scenarios, capable of launching clean, isolated browser sessions in milliseconds. It is redesigned to address Chrome's bottlenecks in batch automation (high memory and CPU usage, slow cold starts, and state crosstalk between tasks), with startup times under 50 milliseconds, each session being a brand-new sandbox that does not share anything with others, memory usage far lower than Chrome, and the ability to run more concurrent sessions on a single machine. It is compatible with Playwright and the Chrome DevTools Protocol (CDP), and existing Puppeteer or Playwright scripts can point to Obscura without rewriting. It offers two approaches: a managed cloud service (a single API key is enough to launch isolated browsers, with no need to build your own servers) and open-source self-hosting (a single Docker container), with no lock-in. It is suitable for large-scale parallel web scraping and AI agent automation tasks.

Open website

About this tool

Obscura is an open-source headless browser engine designed for AI agents and web scraping. Its core goal is to address Chrome's bottlenecks in large-scale automation scenarios: Chrome was built for humans browsing page by page, not for software running thousands of sessions simultaneously, which leads to cost, latency, and unstable results.

Main capabilities

Obscura retains the features machines need, claiming to be faster, cheaper, and safer than Chrome, with no need to rewrite code. Specific features include:

  • Millisecond-level startup: session startup time is under 50 milliseconds, and sessions can be created on demand and destroyed immediately after use.
  • Clean isolation every time: each session is a brand-new sandbox that shares nothing, with no cookie leakage, no contamination, and fully reproducible runs.
  • Higher density: memory usage is only a small fraction of Chrome's, and a single machine can host more concurrent sessions.
  • Preserve existing tech stacks: compatible with Playwright and the Chrome DevTools Protocol (CDP), and existing scripts can point directly to Obscura.

How to use

According to the material, Obscura offers two ways to use it:

  1. Managed cloud service: a single API key is enough to launch isolated browsers in milliseconds, with no need to run your own servers. Navigate, click, type, and extract data through Playwright or CDP without rewriting code; run thousands of sessions in parallel, with capacity configured according to task scale and operated by the official team.
  2. Self-hosting: Obscura is open source, and the engine can be self-hosted in a single Docker container, with no lock-in.

Keep your Playwright or Puppeteer code and point it to Obscura via CDP to connect, drive, and scale.

Use cases

  • Automation tasks where AI agents need independent browsers, with each agent assigned its own browser.
  • Large-scale web scraping: run thousands of parallel sessions, suitable for data collection pipelines.
  • Tasks requiring reproducible results: each session is a brand-new sandbox, avoiding state crosstalk.

Limitations and missing information

The material does not provide details such as specific pricing, plans, performance benchmark data, supported language or platform scope, or open-source license type, which are not yet available. The material also mentions that you can schedule a short call, explain your task, and the official team will demonstrate how Obscura runs and delivers results.