computer-use-agents

The fundamental architecture of computer use agents: observe screen, reason about next action, execute action, repeat. This loop integrates vision models with action execution through an iterative pipeline.

Quality

48%

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

Fix and improve this skill with Tessl

tessl review fix ./plugins/AI-Agents-Safe-Coding-Skills-claude/skills/computer-use-agents/SKILL.md

The canonical home for this skill is computer-use-agents in sickn33/agentic-awesome-skills

Quality

Content

42%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 body presents relevant patterns with real code and config, but the code blocks are truncated and non-executable, the Sharp Edges table is unpopulated with placeholder 'Issue' cells, and everything is inlined with no progressive disclosure or validation checkpoints for destructive desktop operations.

Suggestions

Complete the truncated code blocks so every example is copy-paste executable, or explicitly mark intentionally abbreviated snippets.

Fill the 'Sharp Edges' table with real issue names instead of literal 'Issue' placeholders, and pair each with its concrete solution.

Add explicit validation/feedback checkpoints (e.g., screenshot-diff verification after actions) and move large Dockerfile/compose blocks into referenced bundle files under references/ or scripts/.

Dimension	Reasoning	Score
Conciseness	Mostly code-driven and reasonably efficient, but it includes unnecessary commentary Claude already knows (e.g., 'Claude Opus 4.5 is now the best model in the world for computer use') and time-sensitive version notes that pad the body.	3 / 5
Actionability	Multiple code blocks are truncated mid-line ('scrolled {dir', 'from dataclasses im', an unfinished open() call), so the examples are neither executable nor copy-paste ready, leaving only high-level hints.	2 / 5
Workflow Clarity	The perceive-reason-act loop lists a sequenced 4-step process and a FEEDBACK step is mentioned, but there are no explicit validation checkpoints for these destructive/batch desktop operations, and the 'Sharp Edges' table uses literal 'Issue' placeholders.	3 / 5
Progressive Disclosure	Section headers provide some structure, but no bundle files exist and all code/config (Dockerfile, docker-compose, Python classes) is inlined in one file rather than split into referenced files.	3 / 5
	Total	11 / 20 Passed

Description

53%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 conveys what the skill is about with a recognizable domain term, but it is framed as architecture rather than capability and omits any explicit 'Use when' trigger guidance, capping completeness at 3.

Suggestions

Add an explicit 'Use when...' clause naming concrete triggers (e.g., 'Use when building computer use / desktop automation agents or integrating vision models with screen control').

Reframe around concrete capabilities (e.g., 'drives screenshots, mouse/keyboard actions, and sandboxed desktop control') rather than the abstract perceive-reason-act loop.

Include natural synonyms and extensions users might say ('desktop automation', 'vision agents', 'Anthropic computer use') to improve trigger term coverage.

Dimension	Reasoning	Score
Specificity	It names the domain and a few concrete loop actions ('observe screen, reason about next action, execute action, repeat'), but these are architectural phases rather than task-level capabilities, leaving coverage non-comprehensive.	3 / 5
Completeness	The 'what' is stated clearly (the perceive-reason-act loop), but there is no 'Use when...' or equivalent explicit 'when' guidance, which caps completeness at 3 per the guidelines.	3 / 5
Trigger Term Quality	'computer use agents' is a natural term users would say, but no synonyms or related natural phrases (e.g., 'desktop automation', 'vision agents') are included.	3 / 5
Distinctiveness Conflict Risk	'computer use agents' is a fairly distinct niche with minimal overlap risk against other skills, though the abstract framing leaves minor ambiguity.	4 / 5
	Total	13 / 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: administrakt0r/AI-Agents-Safe-Coding-Skills
Path: plugins/AI-Agents-Safe-Coding-Skills-claude/skills/computer-use-agents/SKILL.md
Commit: 57c135b

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.