browser-use

Control a real browser to navigate pages, click, type, fill forms, inspect rendered UI, take screenshots, or record video. Load only when the user asks to open or automate a browser, interact with or test rendered page UI, scrape a site that needs browser execution, or capture a browser screenshot or video. Do not load for backend logs, traces, API or stream events, source-code inspection, or plain HTTP or web research that does not require a browser.

Quality

96%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

92%Weight 40%Scale 1-5

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

The content is a strong, executable guide to raw CDP automation: actionable code, a sequenced workflow with verification checkpoints, and clean progressive disclosure via a single real reference file. The only minor weakness is length that could be tightened in a few prose passages.

Suggestions

Tighten the 'Launching' and 'Inspecting the page' prose where it restates CDP mechanics already shown in code, to lift conciseness toward the lean anchor.

Consider moving the 'Common failure modes' list into a short troubleshooting reference if the skill grows further, keeping the main body focused on the happy path.

The 'Visible by default when a display exists' section is somewhat discursive; condensing its rationale would reduce tokens without losing the key headless-vs-headful decision rule.

Dimension	Reasoning	Score
Conciseness	The body is dense and protocol-specific (CDP gotchas Claude genuinely needs) with executable code and no padding about basic concepts, but at ~320 lines some explanatory prose around launch/inspection could be trimmed slightly, sitting just below the lean score-5 anchor.	4 / 5
Actionability	Provides copy-paste-ready TypeScript for the common cases — launching, connecting via WebSocket, inspecting, clicking, typing, waiting, and screenshots — fully executable with the listed runtime (Bun/Node ≥22).	5 / 5
Workflow Clarity	A numbered 9-step Workflow sequences find→launch→connect→inspect→act→wait→verify→cleanup with explicit validation ('Verify the result', 'Wait on observable state, never fixed sleeps') and a Debugging-failures section with feedback loops for error recovery.	5 / 5
Progressive Disclosure	SKILL.md is a well-organized overview with section headers, and the only offloaded detail — video recording/demo polish — is a clearly signaled one-level-deep reference to references/recording.md, which exists and is appropriately scoped.	5 / 5
	Total	19 / 20 Passed

Description

100%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.

The description is exemplary: it states concrete capabilities, gives explicit positive and negative load triggers in natural language, and draws a clear boundary against plain HTTP/web-research skills. No vague fluff or over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'navigate pages, click, type, fill forms, inspect rendered UI, take screenshots, or record video' — covering the domain comprehensively, matching the score-5 anchor.	5 / 5
Completeness	Explicitly answers both 'what' (control a real browser to navigate/click/type/fill/inspect/screenshot/record) and 'when' ('Load only when the user asks to...') with concrete trigger phrases and a negative-boundary clause.	5 / 5
Trigger Term Quality	Natural trigger phrases users would say are covered with synonyms — 'open or automate a browser', 'interact with or test rendered page UI', 'scrape a site', 'capture a browser screenshot or video' — plus negative triggers, matching comprehensive coverage.	5 / 5
Distinctiveness Conflict Risk	Clear niche (real browser automation) with distinct triggers and an explicit exclusion list ('Do not load for backend logs, traces, API or stream events, source-code inspection, or plain HTTP'), minimizing conflict with adjacent skills.	5 / 5
	Total	20 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: letta-ai/letta-code
Path: src/skills/builtin/browser-use/SKILL.md
Commit: 7a4337e

Reviewed: 2 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.