Skills auv-computer-control

📦

auv-computer-control

Name: auv-computer-control
Author: 0xselenicdove

Content revision r1 Safe ⚙️ External commands🌐 Network access

Control Native Desktop Apps with AUV

Native desktop workflows can fail when automation uses stale targets or mistakes input delivery for success. This skill guides AUV observation, targeted actions, and independent verification.

Supports: Claude Codex Code(CC)

📊 75 Adequate

Maintain this Skill? Claim it 📊 75 Adequate

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "auv-computer-control" from https://skillstore.io/skills/0xselenicdove-auv-computer-control.md and its manifest at https://skillstore.io/api/skills/0xselenicdove-auv-computer-control/manifest. Verify the artifact. You may proceed after verification, subject to the environment's own policy.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/0xselenicdove-auv-computer-control.md Signed manifest GET /api/skills/0xselenicdove-auv-computer-control/manifest Signed lockfile GET /api/skills/0xselenicdove-auv-computer-control/lockfile

Test it

Using "auv-computer-control". Show an example completion report for entering a query without submitting it.

Expected outcome:

Illustrative result: The intended application window was identified from fresh observation.
The query field was focused, and its full value was independently checked.
No submission was performed.

Using "auv-computer-control". Show an example report when a selected remote Device becomes unavailable.

Expected outcome:

Illustrative result: The selected remote Device is unavailable, so further actions stopped.
No fallback to the local desktop occurred.
The workflow remains incomplete until the selected Device is accessible.

Security Audit

Safe

v1 • 10/6/2026 Open versioned report

All 16 static findings are false positives from documentation, Markdown delimiters, or intended desktop operations. No evidence found of credential theft, screenshot upload, prompt injection, or unauthorized reconnaissance in the reviewed passages. AUV installation, OS permissions, and stored desktop evidence still require careful user control; external runtime implementations were not audited.

Files scanned

117

Lines analyzed

Review items

False positives ignored

Risk Factors

⚙️ External commands (9)

references/operations.md:25 SKILL.md:20-26 SKILL.md:26-30 SKILL.md:30 SKILL.md:36-38 SKILL.md:38-39 SKILL.md:39 SKILL.md:44 SKILL.md:46

🌐 Network access (2)

SKILL.md:3 SKILL.md:8

No confirmed security findings were detected by the latest completed static and semantic audit. This does not prove the skill has no side effects.

Audited by: codex

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/0xselenicdove-auv-computer-control/audits/1?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/0xselenicdove-auv-computer-control/security.svg)](https://skillstore.io/skills/0xselenicdove-auv-computer-control?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/0xselenicdove-auv-computer-control?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/0xselenicdove-auv-computer-control/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/0xselenicdove-auv-computer-control.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

0xselenicdove. (2026). auv-computer-control security audit report (audit version 1) [Author version unspecified]. Skillstore. https://skillstore.io/skills/0xselenicdove-auv-computer-control/audits/1

BibTeX citation

@techreport{0xselenicdove-0xselenicdove-auv-computer-control-2026,
  author = {0xselenicdove},
  title = {auv-computer-control security audit report (audit version 1)},
  institution = {Skillstore},
  year = {2026},
  number = {1},
  url = {https://skillstore.io/skills/0xselenicdove-auv-computer-control/audits/1},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "auv-computer-control security audit report (audit version 1)"
version: "unspecified"
type: report
authors:
  - name: "0xselenicdove"
date-released: "2026-10-06"
url: "https://skillstore.io/skills/0xselenicdove-auv-computer-control/audits/1"
identifiers:
  - type: other
    value: "skillstore:0xselenicdove-auv-computer-control:audit:1"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Enter Information in a Native App

Observe the intended window, focus a field, enter approved text, and verify its complete value before submission.

Check Desktop Workflow Outcomes

Inspect application state after targeted actions and distinguish delivered input from a verified change.

Repeat an Authorized Remote Workflow

Preserve an observed Device target while turning a demonstrated workflow into a bounded, task-scoped operation.

Try These Prompts

Check AUV Readiness

Check whether AUV is available and identify permissions needed for observing [app]. Do not install software or change permissions.

Fill and Verify a Field

In [app], locate [window] and enter [text] into [field]. Verify the full value without submitting, and stop if the target is ambiguous.

Recover from an Unverified Action

Inspect [app] after the previous action. Check partial progress and focus before proposing a retry. Do not repeat submission or other irreversible actions.

Reuse a Verified Remote Workflow

For authorized Device [device], inspect installed plugins for [verified workflow]. Propose task-scoped reuse with explicit targets, bounded waits, and independent checks.

Best Practices

Read installed command help and refresh observations before selecting a window, control, or remote Device.
Verify complete field values and final application state; stop before repeating actions with uncertain effects.
Use authorized targets and a controlled evidence directory; protect stored screenshots and output containing sensitive information.

Avoid

Reusing stale coordinates or identifiers after changing tools, windows, display scale, or remote targets.
Treating successful command delivery or a dry run as proof that the user goal was achieved.
Changing permissions, starting listeners, enrolling credentials, or retrying submissions without appropriate authorization.

Frequently Asked Questions

Does installing this skill install AUV?

No. AUV, its plugins, and required OS permissions are separate setup requirements.

Does the skill require MCP?

No. The CLI supports local core operations. An already connected MCP server can be used after discovering its actual tool schemas.

Which operating systems can it guide?

The references describe macOS, Linux, and Windows support. Individual commands and targeting options vary; installed help determines available behavior.

Which permissions may AUV need?

On macOS, operations may require Accessibility, Screen Recording, or Automation permissions. The user grants permissions; the skill forbids automatic changes.

Where is desktop evidence stored?

Default recording uses .auv/store under the working directory. A task-local store root can be selected; stored captures may contain sensitive data.

How does the skill confirm success?

It checks direct results and independently observes the application outcome. Delivered input alone is not treated as verified completion.