only-cli/oc

Turn any website into a compact CLI tailored for AI agents. Browse the web in hundreds of tokens, not tens of thousands.

What it solves

only-cli/oc provides a way for AI agents to browse the web without consuming excessive tokens. Raw HTML is typically too large for LLM context windows; this tool transforms websites into a compact, numbered command-line interface (CLI) that agents can navigate using simple commands like do, find, and read.

How it works

The tool fetches a web page and renders it as a distilled, numbered view of its content. It uses impers to impersonate real browsers (Chrome or Firefox) to bypass basic bot detection. It maintains session state locally, allowing agents to follow numbered links (e.g., oc do 3) without needing to handle raw URLs. For specific high-traffic sites (like GitHub, Reddit, or Wikipedia), it includes tuned shortcuts that leverage APIs or feeds to provide even more efficient data retrieval.

Who it’s for

Developers building AI agents (such as those using Claude Code, Cursor, or Copilot) who want to give their agents the efficient, token-saving web browsing capabilities.

Highlights

  • Token Efficiency: Reduces tens of thousands of tokens of HTML markup to a few hundred tokens of distilled text.
  • Browser Impersonation: Bypasses naive fetchers by mimicking Chrome and Firefox fingerprints.
  • Site Shortcuts: Built-in optimized paths for dozens of popular documentation sites, forums, and APIs.
  • Session Management: Supports authenticated browsing via cookie seeding and session persistence.
  • Agent-Ready: Includes a dedicated skill for easy integration into agent frameworks.

Related

  • Project
    kunchenguid/chrome-devtools-axi

    chrome‑devtools‑axi is an AXI‑compliant CLI that wraps the Chrome DevTools MCP server, giving LLM agents a low‑token, persistent, and ergonomically‑designed way to drive headless Chrome (navigate, click, fill forms, run JS, take screenshots, run Lighthouse, etc.). It keeps a bridge process alive across commands, uses TOON encoding to cut token usage ~40 %, and ships as an Agent Skill that agents can load on‑demand. Benchmarks show it uses the fewest input tokens, lowest cost, shortest duration, and fewest agent turns among seven browser‑automation setups, with 100 % task success.

  • Project
    jackwener/OpenCLI

    A browser automation CLI that converts websites into deterministic interfaces for humans and AI agents, allowing them to operate websites using logged-in Chrome sessions.

  • Project
    sleepinginsummer/agent-browser-cli

    A CLI tool that connects AI Agents to a real Chrome browser session, allowing them to reuse login states and perform precise page control and perception.

  • Project
    Bin-Huang/camoufox-cli

    A command-line interface and set of skills for AI agents to browse the web using an anti-detect browser that spoofs fingerprints to avoid bot detection.

  • Project
    lexmount/moli

    Moli is an open‑source, Rust‑based headless browser built for AI agents. It provides a full web runtime (HTML, CSS, JavaScript, networking, storage) but keeps memory and CPU low by performing layout and painting only on demand. The same binary can be used via a simple CLI, Chrome DevTools Protocol, or WebDriver Classic/BiDi, letting existing automation tools (Playwright, Selenium, etc.) connect without extra adapters. Benchmarks show Moli uses ~15 % of Chrome’s CPU time and ~13 % of its memory while achieving comparable success rates on real‑world web tasks. It is production‑ready, cross‑platform, and dual‑licensed (Apache‑2.0 or MIT).