# AI use cases for Bladebro (dondai44423/bladebro) Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Use cases: 5 | Page: https://dedupedai.com/use-cases/dondai44423-bladebro/ Bladebro is a Rust browser driver that gives an AI agent full control of a real browser through five MCP tools. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Collect a whole infinite-scroll feed in one call: https://dedupedai.com/use-cases/dondai44423-bladebro/collect-infinite-scroll-feed/ - Run a multi-step web workflow in one agent call: https://dedupedai.com/use-cases/dondai44423-bladebro/run-multi-step-web-workflow/ - Let an agent use your own logged-in browser: https://dedupedai.com/use-cases/dondai44423-bladebro/drive-your-own-logged-in-browser/ - Give Hermes Agent a browser with five tools: https://dedupedai.com/use-cases/dondai44423-bladebro/give-hermes-agent-a-browser/ - Load bot-protected pages that block default browsers: https://dedupedai.com/use-cases/dondai44423-bladebro/browse-bot-protected-sites-with-stealth/ ## How to collect a whole infinite-scroll feed in one call with Bladebro Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README An agent scrolls, dedupes and returns a full feed as one artifact. **Who it is for:** A researcher tracking a long feed or listing that loads more items as you scroll. **The problem:** Reading a feed by hand or by repeated agent calls is slow, costs many tokens, and often produces duplicate items. **The outcome:** The agent returns the collected feed as one deduplicated artifact from a single collect call. ### Steps 1. Run `npm install -g bladebro` to install the CLI and binary. 2. Install Chrome or Chromium, then run `bladebro doctor` to check browser prerequisites. 3. Run `bladebro mcp` and register the server with your AI agent. 4. Ask the agent to open the feed URL and use `act collect` until the feed is exhausted. ### Prompt to paste into your agent ```text Use the bladebro MCP server to open and run act collect until the feed stops loading new items. Return the collected items as one artifact. Verify that no item appears twice in the output. ``` ### Caveat Collecting content from sites may conflict with their terms, so check that you are allowed to collect the feed before running it. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/dondai44423-bladebro/collect-infinite-scroll-feed/ ## How to run a multi-step web workflow in one agent call with Bladebro Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Batch browser actions and branch on page state without a round trip per step. **Who it is for:** An operations person who repeats the same multi-step web task, such as filling a form or checking a set of pages. **The problem:** Driving each click and field entry as a separate agent call is slow and burns tokens on every step. **The outcome:** The agent runs the whole sequence in one batch or run call and reports what changed on the page. ### Steps 1. Run `npm install -g bladebro`. 2. Install Chrome or Chromium on the machine. 3. Run `bladebro mcp` and connect it to your AI agent. 4. Ask the agent to use `act batch` for fixed steps, or `run` when the workflow needs if or while logic. ### Prompt to paste into your agent ```text Use the bladebro MCP server to complete the workflow described in on using act batch or run. Return the final page state and the list of actions taken. Verify the outcome verdict for each step before reporting success. ``` ### Caveat If a call is cancelled, a browser action may still finish, so check the destination page before repeating a mutating step. Risk, credentials: A workflow that submits forms or account changes runs without a human check between steps, so review the steps before letting the agent run them. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/dondai44423-bladebro/run-multi-step-web-workflow/ ## How to let an agent use your own logged-in browser with Bladebro Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Point the agent at your real Chromium-family browser with your profile and logins. **Who it is for:** A user who needs an agent to act inside a site they are already logged into in their own browser. **The problem:** A fresh automated browser has no logins, so the agent cannot reach pages that need your account. **The outcome:** The agent works in your own browser profile, with your logins and display, and reports each action it takes. ### Steps 1. Run `npm install -g bladebro`. 2. Run `bladebro rb on` to switch the browser surfaces to your own Chromium-family browser. 3. Run `bladebro mcp` and connect the server to your AI agent. 4. Ask the agent to open the target site and perform the task, then confirm the result on screen. ### Prompt to paste into your agent ```text Use the bladebro MCP server in real-browser mode to open and complete . Return a summary of each action and the final page state. Verify that the account shown is the one you intended before any change is saved. ``` ### Caveat Real-browser mode turns off the page-injection layer, so it relies on your own real environment rather than the stealth patches. Risk, credentials: The agent acts inside your logged-in profile and can reach your accounts, so confirm each sensitive action before it runs. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/dondai44423-bladebro/drive-your-own-logged-in-browser/ ## How to give Hermes Agent a browser with five tools with Bladebro Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Difficulty: Advanced | Time: 45 minutes | Basis: From the README Connect Hermes Agent to Bladebro's browser tools with one setup command. **Who it is for:** A Hermes Agent user who wants a browser with fewer tool definitions than the built-in one. **The problem:** Hermes' built-in browser toolset is large, and setting up a separate browser for it takes manual config work. **The outcome:** Hermes Agent uses Bladebro's five browser tools after a new chat starts, and the built-in browser toolset is removed from the agent. ### Steps 1. Run `npm install -g bladebro`. 2. Run `bladebro hermes on` to connect the MCP server and switch the browser setup. 3. Start a new Hermes chat, or restart a running gateway or desktop client. 4. Run `bladebro hermes status` to confirm the setup. ### Prompt to paste into your agent ```text In Hermes Agent with Bladebro enabled, open using the browser tools and report the page title and main content. Verify with `bladebro hermes status` that the setup is active before you start. ``` ### Caveat This needs a current Hermes CLI with the config and mcp test commands, and `bladebro hermes off` restores the previous browser settings. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/dondai44423-bladebro/give-hermes-agent-a-browser/ ## How to load bot-protected pages that block default browsers with Bladebro Repo: [dondai44423/bladebro](https://github.com/dondai44423/bladebro) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Use Bladebro's stealth layer to load sites with anti-bot protection. **Who it is for:** An analyst whose agent gets blocked or served challenge pages on sites they have permission to read. **The problem:** Anti-bot systems block automated browsers, so the agent gets no usable content from protected pages. **The outcome:** The agent loads more protected pages and returns the content, with the stealth layer on by default. ### Steps 1. Run `npm install -g bladebro`. 2. Install Chrome or Chromium on the machine. 3. Run `bladebro mcp` and connect it to your AI agent. 4. Ask the agent to open the target page and report whether the content loaded or was blocked. ### Prompt to paste into your agent ```text Use the bladebro MCP server to open and return the main page content. State whether a challenge or block appeared, and verify that the content matches what a normal visitor would see. ``` ### Caveat In the published Stealth Bench V1 run Bladebro cleared 68 of 80 sites, so some protected pages will still block the agent. Risk, reverse engineering: Stealth tooling is built to get past anti-bot defenses, so only use it on sites where your access is permitted. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/dondai44423-bladebro/browse-bot-protected-sites-with-stealth/