agent-browser

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.

2.62x

Quality

88%

Does it follow best practices?

Impact

92%

2.62x

Average score across 6 eval scenarios

Securityby

Critical

Do not install without reviewing

The canonical home for this skill is agent-browser in vercel-labs/agent-browser

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-sequenced browser-automation skill with strong verification loops and a good reference table, but it is longer than necessary for the core file and ships dead references to missing template scripts.

Suggestions

Move niche sections (Experimental: Native Mode, Browser Engine Selection/lightpanda, iOS Simulator) into reference files so the core SKILL.md stays lean for the common case.

The 'Ready-to-Use Templates' section references templates/form-automation.sh, templates/authenticated-session.sh, and templates/capture-workflow.sh, none of which exist in the bundle — either ship those scripts or remove the section to avoid dead references.

Trim duplication between the inline 'Essential Commands'/'Common Patterns' blocks and references/commands.md by keeping only the most essential commands in the body and deferring the rest.

Dimension	Reasoning	Score
Conciseness	The body is ~535 lines of dense, executable command reference with terse comments and almost no conceptual padding, but it keeps several niche sections (Experimental: Native Mode, Browser Engine Selection, iOS Simulator) inline alongside a references/ system, so it could be tightened.	2 / 3
Actionability	Nearly the entire body is copy-paste-ready `agent-browser` commands with concrete flags, expected output, and worked patterns like the form-submission and eval --stdin heredoc examples.	3 / 3
Workflow Clarity	The Core Workflow is a clear 4-step Navigate->Snapshot->Interact->Re-snapshot sequence, reinforced by explicit verification checkpoints (Re-snapshot after DOM changes, 'snapshot -> action -> diff') and a dedicated Ref Lifecycle checklist.	3 / 3
Progressive Disclosure	The Deep-Dive Documentation table signals 7 real one-level-deep reference files with a 'When to Use' column, but the 'Ready-to-Use Templates' section references templates/*.sh scripts that do not exist in the bundle, and sizable inline command/pattern blocks duplicate references/commands.md.	2 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

An exemplary description: third-person, concrete actions, explicit Use-when triggers with natural user phrasings, and a distinct niche that minimizes conflict risk.

Dimension	Reasoning	Score
Specificity	It lists multiple concrete actions — 'navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps' — rather than vague language.	3 / 3
Completeness	It explicitly answers both what ('Browser automation CLI for AI agents') and when ('Use when the user needs to interact with websites... Triggers include...'), with an explicit Use-when clause.	3 / 3
Trigger Term Quality	It provides broad coverage of natural phrasings users would say ('open a website', 'fill out a form', 'take a screenshot', 'scrape data from a page', 'login to a site', 'automate browser actions').	3 / 3
Distinctiveness Conflict Risk	'Browser automation CLI for AI agents' is a clear, distinct niche with specific trigger phrases unlikely to collide with unrelated skills.	3 / 3
	Total	12 / 12 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (540 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
relative_links	Relative link issues: 3 missing	Warning

	Total	13 / 16 Passed

Repository: Intelligent-Internet/ii-agent
Path: src/ii_agent/agents/skills/builtin/agent-browser/SKILL.md
Commit: 0e57985

Reviewed: about 15 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.