agentsSource-backedReview first Safety ✓ Privacy ✓
Computer Use GUI Test Agent
Community reusable agent prompt for end-to-end GUI validation with Claude Code computer use using official documentation: enable the computer-use MCP server, per-app session approval, screenshot checkpoints, and documented example workflows for native apps.
HarnessClaude Code
Review first — review before installing
Open the source and read safety notes before installing.
Safety notes
- Computer use controls real desktop apps with per-app session approval—not sandboxed Bash.
- Review sentinel warnings before approving Terminal, Finder, or System Settings access.
- Press Esc or Ctrl+C to stop computer use and release the session lock immediately.
- Only one Claude Code session can hold the computer-use lock at a time.
Privacy notes
- Screenshots are downscaled before model upload but may still capture sensitive UI content.
- Your terminal window is excluded from screenshots per official documentation.
- Redact customer or proprietary UI details before sharing screenshot evidence externally.
Prerequisites
- Interactive Claude Code session on macOS with computer use enabled per official docs.
- computer-use MCP server enabled through the /mcp menu for the project.
- macOS Accessibility and Screen Recording permissions granted when prompted.
- Target app or simulator approved for the current session when Claude requests control.
Schema details
- Install type
- copy
- Troubleshooting
- No
Source repository stats
- Scope
- Source repo
Full copyable content
## Content
Computer Use GUI Test Agent is a community-authored reusable prompt for GUI validation
with Claude Code computer use. It applies official computer use documentation—not an
official Anthropic QA product.
## Scope Note
This prompt operationalizes documented computer use enablement, approval, and example
workflows from code.claude.com. Release sign-off remains with your team.
## Agent Prompt
You are a computer use GUI validation specialist for Claude Code. Follow official
computer use documentation before attempting GUI work.
Workflow:
1. **Confirm eligibility.** Verify interactive macOS session, supported plan, and computer-use server enabled via /mcp.
2. **Grant permissions.** Ensure Accessibility and Screen Recording permissions are granted when macOS prompts.
3. **Choose a documented pattern.** Pick an official example workflow: validate a native build, reproduce a layout bug, or drive a simulator flow.
4. **Request app approval.** Wait for per-session app approval prompts before controlling target applications.
5. **Execute with screenshots.** Capture screenshot checkpoints at each UI state described in the task, as shown in official examples.
6. **Stop safely.** Use Esc or Ctrl+C if on-screen content looks suspicious or the task completes.
7. **Summarize findings.** Report pass/fail against the requested flow with screenshot references—not a formal test harness.
Output contract:
- Eligibility and enablement checklist results.
- Apps approved during the session.
- Screenshot-indexed observations aligned to the requested workflow.
- Blockers requiring human action (permissions, lock held by another session).
## Features
- Mirrors official computer use enablement and macOS permission steps.
- Applies documented example workflows instead of inventing unsupported QA process.
- Respects per-app session approval and sentinel warnings from the docs.
- Produces screenshot-backed observations using documented downscaling behavior.
## Use Cases
- Build, launch, and click through a native app target after code changes.
- Reproduce a layout bug by resizing windows and screenshotting clipped states.
- Drive an iOS Simulator onboarding flow without writing XCTest first.
- Smoke-test GUI-only tools that lack CLI or MCP integrations.
## Source Notes
Verified against Claude Code computer use documentation on **2026-06-16**:
- Official docs describe enabling the built-in computer-use MCP server from /mcp and
granting macOS Accessibility plus Screen Recording permissions on first use.
- Documentation lists end-to-end UI testing as a primary use case, including opening apps,
clicking through flows, and screenshotting each step without a separate test harness.
- Example workflows cover validating native builds, reproducing layout bugs, and driving
simulator flows with explicit screenshot checkpoints.
- Safety guidance documents per-app session approval, sentinel warnings for high-privilege
apps, Esc to abort, and a single-session lock while computer use is active.
## Duplicate Check
Checked content/agents and content/guides for computer use QA coverage.
computer-use-from-the-claude-code-cli-for-gui-qa is a guides entry for setup.
No agents entry applies official computer use example workflows to a reusable GUI
validation prompt with documented enablement and approval steps.
## Editorial Disclosure
Submitted as an independent community agent entry by kiannidev, based on public Claude
Code computer use documentation and the public anthropics/claude-code repository.
No paid placement, referral, or affiliate relationship.
## Sources
- Claude Code computer use - https://code.claude.com/docs/en/computer-use
- Claude Code MCP - https://code.claude.com/docs/en/mcp
- Claude Code repository - https://github.com/anthropics/claude-codeAbout this resource
Content
Computer Use GUI Test Agent is a community-authored reusable prompt for GUI validation with Claude Code computer use. It applies official computer use documentation—not an official Anthropic QA product.
Scope Note
This prompt operationalizes documented computer use enablement, approval, and example workflows from code.claude.com. Release sign-off remains with your team.
Agent Prompt
You are a computer use GUI validation specialist for Claude Code. Follow official computer use documentation before attempting GUI work.
Workflow:
- Confirm eligibility. Verify interactive macOS session, supported plan, and computer-use server enabled via /mcp.
- Grant permissions. Ensure Accessibility and Screen Recording permissions are granted when macOS prompts.
- Choose a documented pattern. Pick an official example workflow: validate a native build, reproduce a layout bug, or drive a simulator flow.
- Request app approval. Wait for per-session app approval prompts before controlling target applications.
- Execute with screenshots. Capture screenshot checkpoints at each UI state described in the task, as shown in official examples.
- Stop safely. Use Esc or Ctrl+C if on-screen content looks suspicious or the task completes.
- Summarize findings. Report pass/fail against the requested flow with screenshot references—not a formal test harness.
Output contract:
- Eligibility and enablement checklist results.
- Apps approved during the session.
- Screenshot-indexed observations aligned to the requested workflow.
- Blockers requiring human action (permissions, lock held by another session).
Features
- Mirrors official computer use enablement and macOS permission steps.
- Applies documented example workflows instead of inventing unsupported QA process.
- Respects per-app session approval and sentinel warnings from the docs.
- Produces screenshot-backed observations using documented downscaling behavior.
Use Cases
- Build, launch, and click through a native app target after code changes.
- Reproduce a layout bug by resizing windows and screenshotting clipped states.
- Drive an iOS Simulator onboarding flow without writing XCTest first.
- Smoke-test GUI-only tools that lack CLI or MCP integrations.
Source Notes
Verified against Claude Code computer use documentation on 2026-06-16:
- Official docs describe enabling the built-in computer-use MCP server from /mcp and granting macOS Accessibility plus Screen Recording permissions on first use.
- Documentation lists end-to-end UI testing as a primary use case, including opening apps, clicking through flows, and screenshotting each step without a separate test harness.
- Example workflows cover validating native builds, reproducing layout bugs, and driving simulator flows with explicit screenshot checkpoints.
- Safety guidance documents per-app session approval, sentinel warnings for high-privilege apps, Esc to abort, and a single-session lock while computer use is active.
Duplicate Check
Checked content/agents and content/guides for computer use QA coverage. computer-use-from-the-claude-code-cli-for-gui-qa is a guides entry for setup. No agents entry applies official computer use example workflows to a reusable GUI validation prompt with documented enablement and approval steps.
Editorial Disclosure
Submitted as an independent community agent entry by kiannidev, based on public Claude Code computer use documentation and the public anthropics/claude-code repository. No paid placement, referral, or affiliate relationship.
Sources
- Claude Code computer use - https://code.claude.com/docs/en/computer-use
- Claude Code MCP - https://code.claude.com/docs/en/mcp
- Claude Code repository - https://github.com/anthropics/claude-code
Source citations
Add this badge to your README
Show that Computer Use GUI Test Agent is listed on HeyClaude. Paste this Markdown into your README — it renders the badge and links back to this page.
[](https://heyclau.de/entry/agents/computer-use-gui-test-agent)How it compares
Computer Use GUI Test Agent side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
| Field | Computer Use GUI Test Agent Community reusable agent prompt for end-to-end GUI validation with Claude Code computer use using official documentation: enable the computer-use MCP server, per-app session approval, screenshot checkpoints, and documented example workflows for native apps. Open dossier | Dynamic Workflow Migration Planner Agent Community reusable agent prompt for migrating teams to Claude Code dynamic workflows using official workflows documentation: audit phases, YAML or config migration steps, validation gates, and rollback planning for large codebase automation. Open dossier | Agent Observability SRE Agent Community reusable agent prompt for Claude Code analytics and agent platform on-call using official analytics documentation: usage signals, session failure triage, MCP latency patterns, and SRE runbooks for agent hosting teams. Open dossier | AI Agent Cost Governance Analyst Agent Community reusable agent prompt for Claude Code and agent spend governance using official costs documentation: budgets, model tier policy, caching awareness, anomaly triage, and team reporting workflows. Open dossier |
|---|---|---|---|---|
| Trust | ||||
| Install risk | Review first | Review first | Review first | Review first |
| Notes | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ |
| Category | agents | agents | agents | agents |
| Source | source-backed | source-backed | source-backed | source-backed |
| Author | kiannidev | kiannidev | kiannidev | kiannidev |
| Added | 2026-06-16 | 2026-06-16 | 2026-06-16 | 2026-06-16 |
| Platforms | Claude Code | Claude Code | Claude Code | Claude Code |
| Source repo | — | — | — | — |
| Safety notes | ✓Computer use controls real desktop apps with per-app session approval—not sandboxed Bash. Review sentinel warnings before approving Terminal, Finder, or System Settings access. Press Esc or Ctrl+C to stop computer use and release the session lock immediately. Only one Claude Code session can hold the computer-use lock at a time. | ✓Automated workflows can run destructive tools without mid-flight prompts—scope permissions narrowly. Pilot migrations on non-production repositories before enabling org-wide schedules. Rollback plans must include disabling workflow triggers and revoking connector access. Do not migrate production deploy steps until validation gates pass in staging. | ✓Incident commands must not exfiltrate customer prompts into public tickets. Scaling replicas without reviewing tool side effects can amplify destructive MCP calls. Disabling tracing to reduce noise may hide regressions—prefer sampling over full off. Rollback plans should include MCP allowlist and permission settings, not only code. | ✓Cost caps should not push teams toward disabling security controls to save tokens. Investigate MCP or subagent loops before blaming individual users for spikes. Premium model break-glass paths should remain documented for incidents. Governance recommendations require leadership approval before hard enforcement. |
| Privacy notes | ✓Screenshots are downscaled before model upload but may still capture sensitive UI content. Your terminal window is excluded from screenshots per official documentation. Redact customer or proprietary UI details before sharing screenshot evidence externally. | ✓Workflow definitions may embed internal service names, ticket templates, or customer examples. Migration audits can expose legacy prompts with secrets—scrub before archiving in git. Shared workflow repos inherit normal code review and access control requirements. | ✓Analytics and logs may contain prompts, diffs, and credentials if misconfigured. Recommend redaction before exporting incident timelines externally. Shared dashboards should aggregate metrics without raw user content fields. | ✓Cost reports may expose per-user usage; treat exports like sensitive operational data. Do not paste customer content into prompts to debug cost spikes in shared tickets. Aggregate spend in leadership reviews unless investigating an approved incident. |
| Prerequisites |
|
|
|
|
| Install | — | — | — | — |
| Config | — | — | — | — |
| Citations | ||||
| Claim | Unclaimed | Unclaimed | Unclaimed | Unclaimed |
Featured in
Signals
Loading live community signals…
More like this, weekly
A short, calm digest of reviewed Claude resources. Unsubscribe any time.