A well-structured operational guide that assumes Claude's competence and concentrates on genuinely tool-specific behavior (refs, modes, timeouts, popups, multi-batching). Its main weaknesses are run-on passages that could be tightened, prose-only coverage of the core call shapes, and an inline argument catalog that would sit better in a reference file.

Suggestions

Show one complete example call for the core workflow (e.g., the exact tool_args for `open` followed by `content`), mirroring the `multi` JSON example, so the most common path is copy-paste ready.

Break the long Modes paragraph and the evaluate-timeout sentence into short bullets; they each pack 4-5 distinct facts into a single hard-to-parse sentence.

Move the Pointer And Keyboard argument catalog (coordinate args, clipboard, viewport, wheel) into a one-level-deep reference file, keeping SKILL.md to the workflow overview and the most common interactions.

Dimension	Reasoning	Score
Conciseness	The body is dense with tool-specific facts Claude cannot know (browser_id, refs like `[link 3]`, `vision_load.tool_args.paths`, coordinate semantics, evaluate deadline behavior) and explains no general concepts, but a few passages — the long Modes paragraph and the run-on evaluate-timeout/recovery sentence — could be tightened, fitting 'efficient; minor instances of over-explanation' rather than the every-token-earns-its-place anchor 5.	4 / 5
Actionability	Concrete action names, exact argument names, and a copy-paste-ready JSON example for `multi` make the guidance mostly executable, matching anchor 4; it stops short of anchor 5 because the common cases (open → content → click, screenshot) are described in prose without complete example calls showing their full tool_args.	4 / 5
Workflow Clarity	The Core Workflow gives a clear 6-step sequence, and error-recovery feedback is present ("use `list` before retrying a closed tab", the evaluate-timeout reload behavior, "verify files exist" for uploads), matching 'clear sequence with most checkpoints present'; it lacks the explicit validate-before-proceeding checkpoints of anchor 5, and the batch `multi` action carries a caution but no verification step.	4 / 5
Progressive Disclosure	Sections are well organized and the one external reference is clearly signaled twice at one level deep (load `browser-form-workflows` for fragile forms), with no bundle files to inflate the SKILL.md; the Pointer And Keyboard section is a long argument catalog that could arguably move to a separate reference file, keeping this at 'good structure; minor organization gaps' rather than anchor 5.	4 / 5
	Total	16 / 20 Passed

Description

82%Weight 40%Scale 1-5

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

A dense, specific one-line description backed by an unusually thorough explicit triggers list. It clearly states what the skill does and, via the triggers block, when to use it; the only improvement would be folding a 'Use when...' style clause into the description itself and covering a few missing capabilities.

Dimension	Reasoning	Score
Specificity	The description lists several concrete actions ("tabs, clicks, forms, downloads, screenshots; host or Docker browser") rather than vague language, but omits notable capabilities like typing/keyboard input, navigation, and JavaScript execution, leaving minor gaps in coverage — a clear fit for the anchor 'Lists several specific actions; minor gaps' rather than the comprehensive anchor 5.	4 / 5
Completeness	The 'what' is explicit and concrete ("Control web pages: tabs, clicks, forms, downloads, screenshots"), and the 'when' is supplied by the explicit 17-entry triggers block rather than a 'Use when...' clause in the description itself — both present, but the trigger guidance could be more tightly integrated, fitting anchor 4 rather than the fully explicit anchor 5 and clearly above the vague-when anchor 3.	4 / 5
Trigger Term Quality	The frontmatter pairs the description's natural terms ("clicks", "forms", "downloads", "screenshots", "host or Docker browser") with a comprehensive triggers list covering synonyms and phrasing users would actually say ("browser automation", "open website", "open URL", "screenshot webpage", "multi-tab browsing", "upload file in browser", "my browser"), matching the comprehensive-synonyms anchor.	5 / 5
Distinctiveness Conflict Risk	The browser-control niche is clear ("Control web pages... host or Docker browser") and mostly distinct from other skills, but terms like "screenshots", "downloads", and "JavaScript page" carry minor overlap risk with screenshot, download, and web-development skills, matching 'mostly distinct; minor overlap risk' rather than the minimal-conflict anchor 5.	4 / 5
	Total	17 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: agent0ai/agent-zero
Path: plugins/_browser/skills/browser-automation/SKILL.md
Commit: e3051fb

Reviewed: 10 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.