agentleFS
matthewwong525matthewwong525/WongStack/.agents/plugins/browser/SKILL.md

browser

Skill7 starsChanged 4 days agoMITView on GitHub

The person's own browser, for errands.

71 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use V3 Core Implementation, V3 DDD Architecture and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • ruvnetSkill

    V3 Core Implementation

    Core module implementation for claude-flow v3. Implements DDD domains, clean architecture patterns, dependency injection, and modular TypeScript codebase with comprehensive testing.

    Also in 15 projects that use this

  • ruvnetSkill

    V3 DDD Architecture

    Domain-Driven Design architecture for claude-flow v3. Implements modular, bounded context architecture with clean separation of concerns and microkernel pattern.

    Also in 15 projects that use this

  • +10 more from ruvnet

The file

Files
SKILL.md

browser

scripts/browse.mjs drives Camoufox, the browser the agent uses websites in as the person. Browsing owns when and how to run it:

  • open, snapshot, click, type, press, select, get, and screenshot work this task's own tab.
  • A click runs at once, also during another chat's turn. BROWSE_CLICK=queued comes only from an older running browser: run wait-click in the background and take no other step on that page until it ends. BROWSE_FRONT=behind: another chat's turn keeps its tab in front; carry on here.
  • window [on|off] and proxy [on|off] set this computer's browser from its next start: where it shows, the proxy. With the proxy off, a refused site's steps first name turn the browser proxy on.
  • Send me the browser or show me the browser, on every computer: hand-over.mjs open --view --note "Here is the browser. Tap ✅ at the top right when you're done.". It holds the chat's tab as the person's turn. With HANDOVER_VIEW=window, say the 🔔 tab is in the browser window (on a Mac, Camoufox in the Dock if it is behind), and to click ✅ at the top right; on done in the chat, run close. Never front or show for this: neither holds a turn.
  • show brings this task's tab to the front, with no turn, for a finished page only. On BROWSE_SHOWN=link or none, send a picture instead: ask for the browser.
  • A box in another site's frame sits at the end of snapshot: use its ref, not a selector.
  • save and close keep the logins; status, install, and stop manage the browser. status names the route in BROWSE_NETWORK=.
  • Never ask about a password manager: 1Password comes by default. Asked to add any add-on: addon find <name>; for a BROWSE_ADDON_BUILT_IN match, say it is built in and add nothing; for one in addon status, say so. One clear match, featured or 10,000+ daily users: name its maker, website, and daily users, and addon add <slug> at once; else name the matches' facts, with any BROWSE_ADDON_WARN, and add on a yes. Never send them to the add-on site. After an add, tell them its BROWSE_ADDON_TELL; for a password manager, addon open, then send the browser to sign in to it.
  • Sign in with 1Password at every login, before any sign-in turn: addon logins; for a passkey, after the site's passkey button. For each BROWSE_ADDON_LOGIN=<n> | <name> line, passkey: ones too, ask every time, as a multiple choice in the chat, with the question tool when there is one: the login, the site, and your numbered steps once signed in. Only on a yes: addon pick <n>, click sign-in after a fill, and do only those steps; any other step, or outward action, needs its own yes. The end report names the steps done. On locked, none, no password manager, or a no: addon open, then a sign-in turn; the first for 1Password advises an 8-hour auto-lock. A refused pick: addon logins again.
  • An add-on's box over the page, not a passkey box, or a refusal naming one: addon close at once, no ask; report it closed, saving nothing. On its refusal, a live view. Never click, read, type into, or unlock an add-on otherwise.

Preview checks stay on agent-browser.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.