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.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Critical

Do not install without reviewing

Quality

Content

85%

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

The body is highly actionable with executable commands, a clear validated workflow, and a clean one-level-deep reference architecture. Its main weakness is conciseness: substantial cross-section duplication (auth, viewport) inflates length, and three referenced template files are missing.

Suggestions

Consolidate authentication content: the five options in "Handling Authentication" are repeated in "Common Patterns" — keep one canonical treatment and point the other to it to cut length.

Fix or remove the templates section: `templates/form-automation.sh`, `templates/authenticated-session.sh`, and `templates/capture-workflow.sh` are referenced as runnable but do not exist, breaking the copy-paste-ready promise.

Tighten the viewport/device sections, which restate `set viewport` behavior and the `scale` parameter across "Essential Commands" and "Viewport & Responsive Testing".

Dimension	Reasoning	Score
Conciseness	The body is dense and command-driven, but ~640 lines with material repeated across sections — authentication appears as five options in "Handling Authentication" and again in "Common Patterns", and viewport behavior is restated multiple times — so it could be tightened rather than earning the 'every token earns its place' score-3 anchor.	2 / 3
Actionability	Nearly all guidance is concrete, copy-paste-ready bash (e.g. `agent-browser open <url>`, `agent-browser snapshot -i`, `agent-browser fill @e1 "user@example.com"`), matching the fully-executable score-3 anchor; the score-2 pseudocode anchor does not fit.	3 / 3
Workflow Clarity	A clear 4-step Core Workflow (Navigate → Snapshot → Interact → Re-snapshot) is reinforced by the Ref Lifecycle re-snapshot checkpoint and the snapshot→action→diff verification loop, providing explicit validation steps and error-recovery feedback.	3 / 3
Progressive Disclosure	A well-signaled "Deep-Dive Documentation" table links seven real one-level-deep reference files, each with a "When to Use" column, matching the clear-overview score-3 anchor; the only blemish is that the referenced templates/*.sh files do not exist.	3 / 3
	Total	11 / 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.

The description is strong across all dimensions: it states concrete capabilities, provides natural trigger phrases, answers both 'what' and 'when' explicitly, and occupies a distinct niche. It is a model third-person description with no padding.

Dimension	Reasoning	Score
Specificity	Lists multiple specific concrete actions — "navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps" — matching the score-3 anchor rather than the partial score-2 anchor.	3 / 3
Completeness	Explicitly answers both what ("Browser automation CLI for AI agents" plus concrete actions) and when ("Use when the user needs to interact with websites... Triggers include..."), satisfying the score-3 anchor.	3 / 3
Trigger Term Quality	Natural user phrasings like "open a website", "fill out a form", "take a screenshot", "scrape data from a page", "login to a site" give good coverage of terms users would actually say.	3 / 3
Distinctiveness Conflict Risk	A clearly defined browser-automation niche with distinct web-interaction triggers is unlikely to conflict with document, spreadsheet, or other 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 (646 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: databricks/devhub
Commit: 131db95

Reviewed: 5 days 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.