- Home
- Skills
- Workflow Automation
- Runbook AI Browser Agent
Runbook AI Browser Agent
Delegate whole web tasks to an AI agent in your own signed-in Chrome through one MCP tool call. Only the result comes back, and bulky data is written to files, so your context window stays small.
Free
Works with the AI Agents you already use
Runbook AI Browser Agent
Example session with this skill installed
Open https://news.ycombinator.com, read the top 10 stories, and save title, url, points and comment count for each to hn-top10.json.
- Read your context and instructions
- Compiled the runbook ai browser
Result text returned to the coding agent (2 agent iterations, about 15 seconds):
"I saved 10 top Hacker News stories (title, url, points, comment count) to hn-top10.json."
The file hn-top10.json is written to disk and listed by path, not inlined. Its first three entries:
[
{"rank":1,"title":"RIP, Harvest","url":"https://aphyr.com/posts/427-rip-harvest","points":195,"comments":60},
{"rank":2,"title":"A Critique of GPUI","url":"https://blog.zortax.de/posts/a-critique-of-gpui","points":39,"comments":10},
{"rank":3,"title":"KetchUp - Open-source SketchUp alternative built with Rust and WebGPU","url":"https://ketchup3d.github.io/app/","points":339,"comments":92},
...
]
The coding agent's context only received the one-line answer and the file path, not the page.
Connects securely to your tools. The creator never sees your data.
About this skill
Runbook AI is a free Chrome extension with a built-in browser agent. Its MCP server (runbook-ai-mcp, Apache-2.0) exposes one tool, browser-agent. This skill teaches your agent when to use it, how to write the task prompt, and how to read the result.
What it does
- Your agent sends one plain-English task, such as "collect every open invoice on the vendor portal and save them to invoices.json".
- The agent inside the extension navigates, clicks, types and reads pages in your real, signed-in Chrome.
- Only the answer comes back. Scraped data, downloads and screenshots are written to disk and returned as file paths.
When to use it
- Multi-page tasks where a DOM snapshot per click would flood your context.
- Sites that need your existing login: vendor portals, internal tools, LinkedIn, booking sites.
- Sites with no API or MCP server.
Where it stops
- CAPTCHAs and two-factor prompts need you.
- It acts as you in your browser, so the skill tells your agent to confirm before submitting forms, sending messages or buying anything.
- It needs the extension, the MCP server and your own LLM API key. Nothing goes through Runbook AI servers.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 8 days ago
- Free with an Agensi account
Requires