✍️Writing & Content22🎨Image Generation32🎬Video & Animation64🎵Audio & Music48💬Chatbots & Assistants35💻Coding & Development139📈Marketing & SEO52Productivity132🎯Design & UI/UX48📊Data & Analytics31📚Education & Research24💼Business & Finance49🏥Healthcare & Wellness18🔍Search & Knowledge13🤖AI Agent Infrastructure18🛡️AI Security & Testing2🧊3D & Spatial12🔎SEO Tools4🏡Real Estate4🗃️Data Extraction1🧠ADHD & Focus Tools9
Listed in AI Agent Infrastructure with 21 other toolsPart of 891+ curated AI tools on AISO
Libretto logo

Libretto

Open-source browser automation for coding agents — recording CLI, self-healing debug agents, and a low-token Browser Tools SDK

0
freemiumThe CLI and Browser Tools SDK are open source and free (`npm i libretto-browser-tools`). Debug Agents and cloud browsers require an account. No public pricing page exists as of July 2026 — the site routes commercial enquiries through a Talk to a dev link.View full pricing →

Visit Libretto

https://libretto.sh

About Libretto

Libretto builds browser automation tooling for both humans and AI agents, spanning a CLI, an SDK, debugging agents, and cloud browsers. The open-source CLI records website workflows — `npx libretto open` starts a session, captures your actions and the underlying network traffic, and writes them out as fast, reusable Playwright scripts that live in your codebase rather than in a vendor's dashboard. The Debug Agents product handles the part everyone hates: when a Playwright automation breaks, an agent inspects the live page, works out what changed (a renamed field, a moved selector), and opens a pull request with the fix instead of leaving a red build for a human to triage. The Browser Tools SDK is the agent-facing piece — six tools that let any AI agent drive a real browser through Playwright, with adapters for the AI SDK, Pi, and MCP. Its design keeps token cost down: `browser_snapshot` returns a compact accessibility tree with stable refs rather than raw HTML, and `browser_exec` runs Playwright code on the live page and returns only a snapshot diff of what changed. Libretto publishes benchmarks for this — across 26 tasks on public websites, it measured $0.106 per outcome against $0.257 for dev-browser, $0.235 for agent-browser, and $0.293 for playwright-cli, roughly 55% cheaper than the alternatives. It supports Kernel, Browserbase, and local browser providers.

Key Features

Open-source CLI that records browser workflows into reusable Playwright scripts
Network capture alongside action recording
Debug Agents that inspect failing runs and open a PR with the fix
Browser Tools SDK — six tools for agent-driven Playwright control
Compact accessibility-tree snapshots with stable refs instead of raw HTML
Snapshot diffs returned per action to keep token usage low
Adapters for the AI SDK, Pi, and MCP
Works with Kernel, Browserbase, and local browser providers

Libretto Pros & Cons

Pros

  • +Published benchmark showing ~55% lower cost than comparable agent browser harnesses
  • +Core CLI and SDK are open source with no account required
  • +Self-healing PRs address the real maintenance cost of browser automation
  • +Scripts live in your repo, not a vendor dashboard

⚠️ Cons

  • No published pricing for the hosted Debug Agents and cloud browser products
  • Several framework adapters are still marked coming soon
  • Benchmarks are vendor-run, not independently reproduced
  • Assumes Playwright familiarity to get real value from the CLI output

Who Is Libretto Best For?

👤Teams whose Playwright suites break constantly on third-party UI changes
👤Developers giving AI agents controlled access to a real browser
👤Anyone paying too much in tokens for agent browser sessions

Tags

browser automationplaywrightai agentsmcpopen sourcecomputer useweb scraping
🏷️

Is this your tool?

Claim your listing to get a Featured badge, edit your description, and stand out from competitors. All plans include a permanent dofollow backlink to your site.

Claim Now →

ChatGPT already recommends Libretto. Does it recommend yours?

If you're building in AI Agent Infrastructure, run a free AI-visibility scan on your own product — we ask ChatGPT across 5 prompt angles and score how often you get named. ~30 seconds, no signup, no card.

Stay updated on AI Agent Infrastructure tools — join our weekly newsletter

One concise email with fresh launches, trending picks, and featured standouts.

Alternatives to Libretto

View all Libretto alternatives →

Agent connectivity: not yet verified