gemini-computer-use
Build and run Gemini 2.5 Computer Use browser-control agents with Playwright. Use when a user wants to automate web browser tasks via the Gemini Computer Use model, needs an agent loop (screenshot → function_call → action → function_response), or asks to integrate safety confirmation for risky UI actions.
95
93%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
Discovery
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.
This is an excellent skill description that clearly communicates what the skill does (build Gemini 2.5 Computer Use agents with Playwright), when to use it (three explicit trigger scenarios), and uses highly specific terminology that distinguishes it from other automation or AI skills. The description is concise yet comprehensive, with strong trigger terms and a well-structured 'Use when' clause.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: 'Build and run Gemini 2.5 Computer Use browser-control agents with Playwright', describes the agent loop pattern ('screenshot → function_call → action → function_response'), and mentions safety confirmation for risky UI actions. | 3 / 3 |
Completeness | Clearly answers both 'what' (build and run Gemini 2.5 Computer Use browser-control agents with Playwright) and 'when' (explicit 'Use when' clause covering three distinct trigger scenarios: automating browser tasks via Gemini, needing an agent loop, or integrating safety confirmation). | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms users would say: 'Gemini', 'Computer Use', 'browser-control', 'Playwright', 'automate web browser tasks', 'agent loop', 'screenshot', 'safety confirmation', 'risky UI actions'. Good coverage of both high-level and specific terms. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive — the combination of 'Gemini 2.5 Computer Use', 'Playwright', and the specific agent loop pattern creates a very clear niche that is unlikely to conflict with generic browser automation or other AI model skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
87%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, concise skill that provides clear setup instructions and an actionable quick start. The agent loop workflow is well-described but lacks explicit error handling or validation checkpoints for when things go wrong during the loop. Overall it's a strong skill that efficiently communicates the essential information.
Suggestions
Add error recovery guidance to the core workflow, e.g., what to do if the model returns malformed function_calls, if screenshots fail, or if the agent loops without progress.
Consider adding a brief troubleshooting section or validation step (e.g., 'verify Chromium installed: playwright install --check') to catch common setup failures early.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section is lean and purposeful. No unnecessary explanations of what Gemini is, what Playwright does, or how browser automation works. The browser selection section efficiently covers three options in three bullet points. | 3 / 3 |
Actionability | Provides fully executable bash commands for setup and running the agent, including specific flags (--prompt, --start-url, --turn-limit). The env setup, venv creation, dependency installation, and script invocation are all copy-paste ready. | 3 / 3 |
Workflow Clarity | The core agent loop is clearly sequenced with 6 numbered steps including the safety confirmation checkpoint. However, there's no explicit validation or error recovery guidance—e.g., what to do if the model returns unexpected responses, if Playwright fails to capture a screenshot, or if the agent gets stuck in a loop. | 2 / 3 |
Progressive Disclosure | Clean overview structure with a Resources section pointing to the script, reference notes, and env template—all one level deep and clearly signaled. Content is appropriately split between the SKILL.md overview and referenced files. | 3 / 3 |
Total | 11 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- am-will/codex-skills
- Commit
633b7e0
- Reviewed
Table of Contents
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.