How do I install the Browser skill?

Run npx skillmds@latest add tomevault-io/browser-7 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Browser safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Browser?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Browser free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Browser?

tomevault-io (@tomevault-io) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Browser by tomevault-io — Agent Skill

Name: Browser
Author: tomevault-io

Browser

Automate browser interactions using the Browserbase browse CLI. Use when a task requires navigating websites, reading page state, clicking elements, filling forms, or validating browser results with the browse command. Use when this capability is needed.

tomevault-io Updated Sep 17, 2026

File contents

Browser Automation

Automate browser interactions using the browse CLI with Claude.

Setup

Do not install packages during evals. The eval harness provides browse on PATH and pins it to the active benchmark session.

Environment Selection

The eval harness pins browse to one isolated session and preselects the intended environment.

Local mode uses local Chrome.
Remote mode uses Browserbase when credentials are configured.
Do not switch environments unless the task or harness explicitly requires it.

Page State

Prefer browse snapshot over screenshots. It returns a structured accessibility tree with element refs that can be used for interactions.

Only use screenshots when visual layout or image content is necessary.

Useful state checks:

browse status
browse get url
browse get title
browse snapshot

Interaction Workflow

Typical flow:

Navigate to the starting URL with browse open <url>.
Run browse snapshot to inspect structure and element refs.
Interact with refs from the snapshot, selectors, or keyboard actions.
Run browse snapshot or another state check to verify the result.
Repeat until the benchmark instruction is complete.

Common interaction patterns:

browse click <ref>
browse type <text>
browse press <key>
browse fill <selector> <value>
browse wait timeout <ms>

Best Practices

Always navigate before interacting.
Use refs from browse snapshot when possible.
Verify each important action before moving on.
Keep actions focused on the benchmark instruction.
Do not edit repository files.
Do not use browser/network tools outside the provided browse command.

Completion

When finished, report the benchmark result in the exact EVAL_RESULT format requested by the harness prompt.

Source: browserbase/stagehand — distributed by TomeVault.

---name: browser-7description: Automate browser interactions using the Browserbase browse CLI. Use when a task requires navigating websites, reading page state, clicking elements, filling forms, or validating browser results with the browse command. Use when this capability is needed.---# Browser AutomationAutomate browser interactions using the browse CLI with Claude.## SetupDo not install packages during evals. The eval harness provides `browse` on `PATH` and pins it to the active benchmark session.## Environment SelectionThe eval harness pins `browse` to one isolated session and preselects the intended environment.- Local mode uses local Chrome.- Remote mode uses Browserbase when credentials are configured.- Do not switch environments unless the task or harness explicitly requires it.## Page StatePrefer `browse snapshot` over screenshots. It returns a structured accessibility tree with element refs that can be used for interactions.Only use screenshots when visual layout or image content is necessary.Useful state checks:```bashbrowse statusbrowse get urlbrowse get titlebrowse snapshot```## Interaction WorkflowTypical flow:1. Navigate to the starting URL with `browse open <url>`.2. Run `browse snapshot` to inspect structure and element refs.3. Interact with refs from the snapshot, selectors, or keyboard actions.4. Run `browse snapshot` or another state check to verify the result.5. Repeat until the benchmark instruction is complete.Common interaction patterns:```bashbrowse click <ref>browse type <text>browse press <key>browse fill <selector> <value>browse wait timeout <ms>```## Best Practices- Always navigate before interacting.- Use refs from `browse snapshot` when possible.- Verify each important action before moving on.- Keep actions focused on the benchmark instruction.- Do not edit repository files.- Do not use browser/network tools outside the provided `browse` command.## CompletionWhen finished, report the benchmark result in the exact `EVAL_RESULT` format requested by the harness prompt.---> Source: [browserbase/stagehand](https://github.com/browserbase/stagehand) — distributed by [TomeVault](https://tomevault.io).

Frequently asked questions