{"id":"896a01c7-8a7a-440b-9545-f56d8d2669df","entityType":"agent","slug":"clawhub-athola-nm-phantom-computer-control","name":"computer-control","canonicalUrl":"https://www.xpersona.co/agent/clawhub-athola-nm-phantom-computer-control","canonicalPath":"/agent/clawhub-athola-nm-phantom-computer-control","generatedAt":"2026-10-10T10:43:32.872Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":null},"description":"Automates desktop GUI workflows via computer use API with screenshot capture Skill: computer-control Owner: athola Summary: Automates desktop GUI workflows via computer use API with screenshot capture Tags: latest:1.9.19 Version history: v1.9.19 | 2026-08-26T13:19:50.470Z | user Release v1.9.19 v1.9.17 | 2026-07-30T05:39:58.522Z | user Release v1.9.17 v1.9.16 | 2026-07-14T19:56:46.419Z | user Release v1.9.16 v1.9.14 | 2026-06-30T18:04:54.047Z | user Release v1.9.14 v1.9.13 | 2026-06-27T16:22:","descriptionLabel":"Technical summary","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1.6K downloads reported by the source. Last updated 10/10/2026.","installCommand":"clawhub skill install s17emme0e2m3cpf7k2jvp3a84984b8z9:nm-phantom-computer-control","sourceUrl":"https://clawhub.ai/athola/nm-phantom-computer-control","homepage":"https://clawhub.ai/athola/skills/nm-phantom-computer-control","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/athola/nm-phantom-computer-control","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/athola/skills/nm-phantom-computer-control","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":64,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Automates desktop GUI workflows via computer use API with screenshot capture Skill: computer-control Owner: athola Summary: Automates desktop GUI workflows via "},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":null},"protocols":[{"protocol":"OPENCLEW","label":"OpenClaw","status":"self-declared","notes":"Declared in the public agent profile."}],"capabilities":[],"verifiedCount":0,"selfDeclaredCount":1,"capabilityMatrix":{"rows":[{"key":"OPENCLEW","type":"protocol","support":"unknown","confidenceSource":"profile","notes":"Listed on profile"}],"flattenedTokens":"protocol:OPENCLEW|unknown|profile"}},"adoption":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":null},"stars":null,"forks":null,"downloads":1648,"packageName":null,"latestVersion":"1.9.19","tractionLabel":"1.6K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":null},"lastUpdatedAt":"2026-10-10T05:37:34.766Z","lastCrawledAt":"2026-10-10T05:37:34.766Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-11T05:37:34.766Z","lastVerifiedAt":null,"highlights":[{"version":"1.9.19","createdAt":"2026-08-26T13:19:50.470Z","changelog":"Release v1.9.19","fileCount":3,"zipByteSize":3896},{"version":"1.9.17","createdAt":"2026-07-30T05:39:58.522Z","changelog":"Release v1.9.17","fileCount":3,"zipByteSize":4098},{"version":"1.9.16","createdAt":"2026-07-14T19:56:46.419Z","changelog":"Release v1.9.16","fileCount":3,"zipByteSize":3974},{"version":"1.9.14","createdAt":"2026-06-30T18:04:54.047Z","changelog":"Release v1.9.14","fileCount":3,"zipByteSize":4038},{"version":"1.9.13","createdAt":"2026-06-27T16:22:47.956Z","changelog":"Release v1.9.13","fileCount":3,"zipByteSize":3998},{"version":"1.9.12","createdAt":"2026-06-19T03:18:06.684Z","changelog":"Release v1.9.12","fileCount":3,"zipByteSize":4028},{"version":"1.0.3","createdAt":"2026-06-18T15:18:05.275Z","changelog":"Release v1.9.12","fileCount":3,"zipByteSize":4052},{"version":"1.0.2","createdAt":"2026-05-09T02:19:38.325Z","changelog":"Release v1.9.5","fileCount":3,"zipByteSize":3899}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s17emme0e2m3cpf7k2jvp3a84984b8z9:nm-phantom-computer-control","setupComplexity":"low","setupSteps":["Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.","Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data."],"contract":{"contractStatus":"missing","authModes":[],"requires":[],"forbidden":[],"supportsMcp":false,"supportsA2a":false,"supportsStreaming":false,"inputSchemaRef":null,"outputSchemaRef":null,"dataRegion":null,"contractUpdatedAt":null,"sourceUpdatedAt":null,"freshnessSeconds":null},"invocationGuide":{"preferredApi":{"snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-10T10:43:32.870Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-athola-nm-phantom-computer-control/trust"}},"reliability":{"evidence":{"source":"runtime-metrics","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No trust, reliability, or runtime telemetry is available."},"trust":{"status":"unavailable","handshakeStatus":"UNKNOWN","verificationFreshnessHours":null,"reputationScore":null,"p95LatencyMs":null,"successRate30d":null,"fallbackRate":null,"attempts30d":null,"trustUpdatedAt":null,"trustConfidence":"unknown","sourceUpdatedAt":null,"freshnessSeconds":null},"decisionGuardrails":{"doNotUseIf":["Contract metadata is missing or unavailable for deterministic execution."],"safeUseWhen":[],"riskFlags":["missing_or_unavailable_contract","trust_data_unavailable","schema_references_missing"],"operationalConfidence":"low"},"executionMetrics":{"observedLatencyMsP50":null,"observedLatencyMsP95":null,"estimatedCostUsd":null,"uptime30d":null,"rateLimitRpm":null,"rateLimitBurst":null,"lastVerifiedAt":null,"verificationSource":null},"runtimeMetrics":{"successRate":null,"avgLatencyMs":null,"avgCostUsd":null,"hallucinationRate":null,"retryRate":null,"disputeRate":null,"p50Latency":null,"p95Latency":null,"lastUpdated":null}},"benchmarks":{"evidence":{"source":"no-benchmark-data","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No benchmark suites or observed failure patterns are available."},"suites":[],"failurePatterns":[]},"artifacts":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"high","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":null},"readme":"Skill: computer-control\n\nOwner: athola\n\nSummary: Automates desktop GUI workflows via computer use API with screenshot capture\n\nTags: latest:1.9.19\n\nVersion history:\n\nv1.9.19 | 2026-08-26T13:19:50.470Z | user\n\nRelease v1.9.19\n\nv1.9.17 | 2026-07-30T05:39:58.522Z | user\n\nRelease v1.9.17\n\nv1.9.16 | 2026-07-14T19:56:46.419Z | user\n\nRelease v1.9.16\n\nv1.9.14 | 2026-06-30T18:04:54.047Z | user\n\nRelease v1.9.14\n\nv1.9.13 | 2026-06-27T16:22:47.956Z | user\n\nRelease v1.9.13\n\nv1.9.12 | 2026-06-19T03:18:06.684Z | user\n\nRelease v1.9.12\n\nv1.0.3 | 2026-06-18T15:18:05.275Z | user\n\nRelease v1.9.12\n\nv1.0.2 | 2026-05-09T02:19:38.325Z | user\n\nRelease v1.9.5\n\nv1.0.1 | 2026-05-06T14:20:56.494Z | user\n\nRelease v1.9.4\n\nv1.0.0 | 2026-04-15T17:01:37.064Z | auto\n\nInitial release of the Computer Control skill.\n\n- Automate desktop GUI workflows using Claude computer use API.\n- Supports screenshot capture, mouse/keyboard control, and visual task automation.\n- CLI and Python interfaces provided for easy integration.\n- Compatibility with Opus/Sonnet/Haiku models and multiple tool versions.\n- Includes environment setup instructions, API version mapping, and safety guidelines.\n\nArchive index:\n\nArchive v1.9.19: 3 files, 3896 bytes\n\nFiles: skill-card.md (1863b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.19:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.19:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.19\",\n  \"publishedAt\": 1787750390470\n}\n\nFile v1.9.19:skill-card.md\n\n## Description:\n\nAutomates desktop GUI workflows via computer use API with screenshot capture.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[athola](https://clawhub.ai/user/athola)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and engineers use this skill to automate and verify GUI workflows that lack CLI or API alternatives, including visual web-app testing, form filling, and desktop app navigation.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: The skill can capture screenshots and synthesize keyboard and mouse input in other desktop applications.\n\nMitigation: Run it in a sandbox, VM, Docker, or controlled desktop environment, and close private windows before use.\n\nRisk: GUI actions can affect sensitive accounts or trigger irreversible real-world changes.\n\nMitigation: Avoid banking and sensitive services, and require explicit human confirmation before consequential actions.\n\n## Reference(s):\n\n- [ClawHub skill page](https://clawhub.ai/athola/skills/nm-phantom-computer-control)\n- [OpenClaw homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom)\n\n## Skill Output:\n\n**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance]\n\n**Output Format:** [Markdown guidance with shell command and Python code examples]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May require desktop screenshot and input access; use iteration caps for cost and runaway-action control.]\n\n## Skill Version(s):\n\n1.9.19 (source: ClawHub release evidence)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.\n\nArchive v1.9.17: 3 files, 4098 bytes\n\nFiles: skill-card.md (2431b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.17:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.17:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.17\",\n  \"publishedAt\": 1785389998522\n}\n\nFile v1.9.17:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and automation engineers use this skill to guide agents through GUI workflows, visual web app testing, form filling, menu navigation, and desktop tasks that require screenshot-based verification. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can capture the visible desktop and may expose private information in screenshots. <br>\nMitigation: Close private applications before use and run the workflow in an isolated desktop, VM, or container when possible. <br>\nRisk: Mouse and keyboard automation can cause cross-application side effects or real-world actions. <br>\nMitigation: Require confirmation before destructive or real-world actions and keep the workflow explicitly opt-in. <br>\nRisk: Long-running GUI loops can consume API budget or continue after the task stops being useful. <br>\nMitigation: Set iteration caps and monitor execution during automation. <br>\nRisk: Supplying sensitive credentials to automated GUI sessions increases account exposure. <br>\nMitigation: Avoid sensitive accounts such as banking and provide login credentials only when necessary. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/skills/nm-phantom-computer-control) <br>\n- [Night Market phantom plugin homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Guidance, Markdown, Code, Shell commands, Configuration instructions] <br>\n**Output Format:** [Markdown with inline bash and Python code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include GUI automation steps, environment checks, API usage examples, and safety guidance.] <br>\n\n## Skill Version(s): <br>\n1.9.17 (source: server release evidence; artifact frontmatter says 1.9.8) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.9.16: 3 files, 3974 bytes\n\nFiles: skill-card.md (2157b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.16:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.16:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.16\",\n  \"publishedAt\": 1784059006419\n}\n\nFile v1.9.16:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and automation engineers use this skill to direct an agent through GUI workflows that require screenshots, mouse and keyboard actions, visual testing, form filling, or desktop-app navigation. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Desktop automation can interact with sensitive accounts, private data, or workflows with real-world consequences. <br>\nMitigation: Run the skill in a sandbox or VM, avoid sensitive accounts and banking, and require human confirmation before consequential actions. <br>\nRisk: Screenshots can expose private information visible on the desktop. <br>\nMitigation: Close sensitive apps and use a limited desktop session before giving the agent screenshot access. <br>\nRisk: Long-running computer-use loops can repeat unwanted actions or increase API cost. <br>\nMitigation: Set iteration limits and stop the session when the task is complete or behavior diverges from the requested workflow. <br>\n\n\n## Reference(s): <br>\n- [Phantom plugin homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [guidance, markdown, code, shell commands, configuration] <br>\n**Output Format:** [Markdown guidance with shell and Python code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include environment setup commands, model and tool-version guidance, and safety constraints.] <br>\n\n## Skill Version(s): <br>\n1.9.16 (source: release evidence; artifact frontmatter says 1.9.8) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.9.14: 3 files, 4038 bytes\n\nFiles: skill-card.md (2305b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.14:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.14:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.14\",\n  \"publishedAt\": 1782842694047\n}\n\nFile v1.9.14:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and engineers use this skill to ask an agent to inspect and control GUI or browser workflows with screenshots, mouse and keyboard actions, and visual verification when CLI or API automation is not enough. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can guide an agent through GUI or browser actions such as typing, clicking, submitting forms, and taking screenshots. <br>\nMitigation: Use it deliberately, require confirmation before consequential actions, and set clear task boundaries before starting automation. <br>\nRisk: Screenshots may expose sensitive windows or private information visible on the desktop. <br>\nMitigation: Close sensitive applications and avoid using the skill with banking, credential, or other sensitive services unless explicitly necessary. <br>\nRisk: A broad activation phrase could start automation in contexts where GUI control is not intended. <br>\nMitigation: Treat GUI automation as opt-in and prefer CLI, API, Playwright, or CDP workflows when those are sufficient. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/skills/nm-phantom-computer-control) <br>\n- [Clawdis homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance] <br>\n**Output Format:** [Markdown with inline code and shell command examples] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include GUI automation steps, environment checks, Python snippets, and safety guidance.] <br>\n\n## Skill Version(s): <br>\n1.9.14 (source: server release evidence) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.9.13: 3 files, 3998 bytes\n\nFiles: skill-card.md (2190b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.13:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.13:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.13\",\n  \"publishedAt\": 1782577367956\n}\n\nFile v1.9.13:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and engineers use this skill to automate visual desktop workflows, test web applications through GUI interaction, fill forms, navigate menus, and verify actions from screenshots when CLI or API paths are not sufficient. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can expose desktop contents through screenshots and control the keyboard and mouse. <br>\nMitigation: Install only when intentional, run it in an isolated VM, container, or display session, and close private applications before use. <br>\nRisk: Automated GUI actions can affect sensitive accounts or produce real-world consequences. <br>\nMitigation: Avoid banking or other sensitive services and require human confirmation before consequential actions. <br>\nRisk: Long-running agent loops can continue taking actions or incur API cost. <br>\nMitigation: Use iteration caps and monitor runs during desktop automation. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/skills/nm-phantom-computer-control) <br>\n- [Project homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Guidance, Shell commands, Code, Configuration] <br>\n**Output Format:** [Markdown with inline bash and Python code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include safety guidance for sandboxing, iteration limits, and human confirmation before consequential GUI actions.] <br>\n\n## Skill Version(s): <br>\n1.9.13 (source: server release metadata) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.9.12: 3 files, 4028 bytes\n\nFiles: skill-card.md (2249b), SKILL.md (4951b), _meta.json (147b)\n\nFile v1.9.12:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.9.12:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.12\",\n  \"publishedAt\": 1781839086684\n}\n\nFile v1.9.12:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and automation engineers use this skill to run opt-in computer-use workflows that inspect GUI state through screenshots and drive mouse or keyboard actions for desktop and browser tasks. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Screenshot capture can expose private information visible on the desktop. <br>\nMitigation: Run the workflow in a sandbox, VM, or controlled display session and close sensitive applications before use. <br>\nRisk: Synthesized mouse and keyboard actions can change application state or trigger real-world effects. <br>\nMitigation: Keep the skill opt-in, use human approval for consequential actions, and prefer dry-run or confirmation steps where available. <br>\nRisk: Long-running computer-use loops can consume API budget or continue acting after a task becomes ambiguous. <br>\nMitigation: Set iteration caps and review progress between steps, especially for workflows involving credentials, package installs, or production systems. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/nm-phantom-computer-control) <br>\n- [Project homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance] <br>\n**Output Format:** [Markdown with inline shell and Python code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include screenshot-informed action plans, environment checks, and confirmation guidance.] <br>\n\n## Skill Version(s): <br>\n1.9.12 (source: server release metadata) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.3: 3 files, 4052 bytes\n\nFiles: skill-card.md (2282b), SKILL.md (4951b), _meta.json (146b)\n\nFile v1.0.3:SKILL.md\n\n---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.0.3:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.0.3\",\n  \"publishedAt\": 1781795885275\n}\n\nFile v1.0.3:skill-card.md\n\n## Description: <br>\nAutomates desktop GUI workflows via computer use API with screenshot capture. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and engineers use this skill to automate GUI workflows that lack CLI alternatives, including browser sessions, form filling, desktop app navigation, visual testing, and screenshot-based verification. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can view a desktop through screenshots and control mouse and keyboard input, which may expose private information or affect other applications. <br>\nMitigation: Run it in a VM, container, or other isolated display when possible, and close private windows or sensitive applications before use. <br>\nRisk: Desktop automation can perform actions with real-world consequences if it is allowed to operate sensitive accounts or services. <br>\nMitigation: Avoid banking and other sensitive accounts, and require human confirmation before consequential actions. <br>\nRisk: Long-running GUI automation can consume time, API budget, or continue beyond the intended task. <br>\nMitigation: Set explicit iteration limits for each run. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/nm-phantom-computer-control) <br>\n- [Project homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Guidance, Shell commands, Code, Configuration] <br>\n**Output Format:** [Markdown with inline bash and Python code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May guide an agent to view screenshots and issue mouse or keyboard actions in a desktop session.] <br>\n\n## Skill Version(s): <br>\n1.0.3 (source: server release metadata; artifact frontmatter lists 1.9.8) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.2: 3 files, 3899 bytes\n\nFiles: skill-card.md (2545b), SKILL.md (4592b), _meta.json (146b)\n\nFile v1.0.2:SKILL.md\n\n---\nname: computer-control\ndescription: |\n  Automate desktop GUI workflows via Claude computer use API with screenshot capture and mouse/keyboard control\nversion: 1.9.5\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.0.2:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.0.2\",\n  \"publishedAt\": 1778293178325\n}\n\nFile v1.0.2:skill-card.md\n\n## Description: <br>\nAutomate desktop GUI workflows via Claude computer use API with screenshot capture and mouse/keyboard control. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[athola](https://clawhub.ai/user/athola) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and engineers use this skill to automate GUI workflows that require visual inspection, form entry, menu navigation, desktop application control, or web application testing through screenshots and mouse or keyboard actions. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill enables an agent to view and control a desktop through screenshots and mouse or keyboard actions. <br>\nMitigation: Run it in a VM, container, or clean desktop and close private applications before use. <br>\nRisk: Desktop screenshots may expose private content or logged-in services to the agent. <br>\nMitigation: Avoid sensitive accounts and do not provide credentials unless they are required for the intended workflow. <br>\nRisk: The skill requires protection of an Anthropic API key. <br>\nMitigation: Store the API key securely and limit access to the execution environment. <br>\nRisk: Automated GUI actions can have unintended real-world consequences. <br>\nMitigation: Use iteration caps and require human approval for consequential actions. <br>\nRisk: The release references external phantom plugin code. <br>\nMitigation: Review the referenced external plugin code before running it. <br>\n\n\n## Reference(s): <br>\n- [ClawHub skill page](https://clawhub.ai/athola/nm-phantom-computer-control) <br>\n- [Phantom plugin homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance] <br>\n**Output Format:** [Markdown with inline shell commands and Python examples] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May include desktop automation steps, environment setup commands, API usage examples, and safety guidance.] <br>\n\n## Skill Version(s): <br>\n1.0.2 (source: server release metadata; artifact frontmatter lists 1.9.5) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.1: 2 files, 2606 bytes\n\nFiles: SKILL.md (4592b), _meta.json (146b)\n\nFile v1.0.1:SKILL.md\n\n---\nname: computer-control\ndescription: |\n  Automate desktop GUI workflows via Claude computer use API with screenshot capture and mouse/keyboard control\nversion: 1.9.4\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.0.1:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.0.1\",\n  \"publishedAt\": 1778077256494\n}\n\nArchive v1.0.0: 2 files, 2606 bytes\n\nFiles: SKILL.md (4592b), _meta.json (146b)\n\nFile v1.0.0:SKILL.md\n\n---\nname: computer-control\ndescription: |\n  Automate desktop GUI workflows via Claude computer use API with screenshot capture and mouse/keyboard control\nversion: 1.8.2\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, Opus 4.5 | `computer_20251124` | `computer-use-2025-11-24` |\n| Sonnet 4.5, Haiku 4.5, older | `computer_20250124` | `computer-use-2025-01-24` |\n\nThe `resolve_tool_version()` function handles this mapping\nautomatically based on the model name.\n\n## Available Actions\n\n**All versions:**\n- `screenshot` - capture display\n- `left_click` - click at `[x, y]`\n- `type` - type text string\n- `key` - press key combo (e.g., `ctrl+s`)\n- `mouse_move` - move cursor\n\n**Enhanced (20250124+):**\n- `scroll` - scroll with direction and amount\n- `left_click_drag` - drag between coordinates\n- `right_click`, `middle_click`, `double_click`, `triple_click`\n- `hold_key` - hold key for duration\n- `wait` - pause between actions\n\n**Latest (20251124):**\n- `zoom` - inspect screen region at full resolution\n\n## Safety\n\nComputer use carries risks. Follow these guidelines:\n\n1. **Use a sandbox**: Run in Docker or a VM, not your main OS\n2. **Limit access**: Do not provide login credentials unless\n   necessary, and never for banking or sensitive services\n3. **Set iteration caps**: Always use `max_iterations` to\n   prevent runaway API costs\n4. **Human approval**: For actions with real-world consequences,\n   add confirmation callbacks via `on_action`\n5. **Close sensitive apps**: Claude sees the full screen via\n   screenshots; close anything private before starting\n\n## Environment Requirements\n\n**Linux (native or WSL2 with WSLg):**\n\n```bash\nsudo apt install xdotool scrot xclip\n```\n\n**Headless (Docker/CI):**\n\n```bash\n# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1\n```\n\n## Prompting Tips\n\n1. Be specific about each step of the task\n2. Add \"After each step, take a screenshot and verify\" to\n   catch mistakes early\n3. Use keyboard shortcuts when UI elements are hard to click\n4. Provide example screenshots for repeatable workflows\n5. Set a system prompt with domain-specific instructions\n\nFile v1.0.0:_meta.json\n\n{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.0.0\",\n  \"publishedAt\": 1776272497064\n}","readmeExcerpt":"Skill: computer-control Owner: athola Summary: Automates desktop GUI workflows via computer use API with screenshot capture Tags: latest:1.9.19 Version history: v1.9.19 | 2026-08-26T13:19:50.470Z | user Release v1.9.19 v1.9.17 | 2026-07-30T05:39:58.522Z | user Release v1.9.17 v1.9.16 | 2026-07-14T19:56:46.419Z | user Release v1.9.16 v1.9.14 | 2026-06-30T18:04:54.047Z | user Release v1.9.14 v1.9.13 | 2026-06-27T16:22:","codeSnippets":[],"executableExamples":[{"language":"text","snippet":"User Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)"},{"language":"bash","snippet":"cd plugins/phantom\nuv run python -m phantom.cli --check"},{"language":"bash","snippet":"export ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\""},{"language":"python","snippet":"from phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)"},{"language":"bash","snippet":"sudo apt install xdotool scrot xclip"},{"language":"bash","snippet":"# Install Xvfb for virtual display\nsudo apt install xvfb xdotool scrot xclip\nXvfb :1 -screen 0 1920x1080x24 &\nexport DISPLAY=:1"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: computer-control\ndescription: Automates desktop GUI workflows via computer use API with screenshot capture\nversion: 1.9.8\ntriggers:\n  - scripting GUI interactions or recording browser sessions for tutorials\nmetadata: {\"openclaw\": {\"homepage\": \"https://github.com/athola/claude-night-market/tree/master/plugins/phantom\", \"emoji\": \"\\ud83e\\udd9e\"}}\nsource: claude-night-market\nsource_plugin: phantom\n---\n\n> **Night Market Skill** — ported from [claude-night-market/phantom](https://github.com/athola/claude-night-market/tree/master/plugins/phantom). For the full experience with agents, hooks, and commands, install the Claude Code plugin.\n\n\n# Computer Control Skill\n\nUse Claude's Computer Use API to see and control desktop\nenvironments through screenshots and mouse/keyboard actions.\n\n## When To Use\n\n- Automating GUI-based workflows that lack CLI alternatives\n- Testing web applications through visual interaction\n- Filling forms, navigating menus, or interacting with desktop apps\n- Building automation pipelines that need visual verification\n\n## When NOT To Use\n\n- Tasks achievable through CLI or API (no GUI needed)\n- Browser automation better served by Playwright or CDP\n\n> **Why this stays opt-in.** Per\n> [docs/inclusive-defaults.md][inc] (TRUE-exception\n> category 4), Computer Use takes screenshots and\n> synthesizes keyboard/mouse input — cross-process side\n> effects that must always be explicitly invoked, never\n> default-on.\n\n[inc]: ../../../../docs/inclusive-defaults.md\n\n## Architecture\n\nThe computer use system has three layers:\n\n1. **Display Toolkit** (`phantom.display`) - executes OS-level\n   actions via xdotool/scrot on the real or virtual display\n2. **Agent Loop** (`phantom.loop`) - manages the conversation\n   cycle between Claude API and the display toolkit\n3. **CLI** (`phantom.cli`) - command-line interface for running\n   tasks or checking environment readiness\n\n```\nUser Task\n    |\n    v\nAgent Loop  <---->  Claude API (beta)\n    |                   |\n    v                   v\nDisplay Toolkit    tool_use responses\n    |              (click, type, screenshot)\n    v\nOS Commands (xdotool, scrot)\n    |\n    v\nDisplay (X11 / Xvfb / WSLg)\n```\n\n## Quick Start\n\n### Check environment\n\n```bash\ncd plugins/phantom\nuv run python -m phantom.cli --check\n```\n\n### Run a task\n\n```bash\nexport ANTHROPIC_API_KEY=\"sk-ant-...\"\nuv run python -m phantom.cli \"Open Firefox and search for Claude AI\"\n```\n\n### Use in Python\n\n```python\nfrom phantom.display import DisplayConfig, DisplayToolkit\nfrom phantom.loop import LoopConfig, run_loop\n\nresult = run_loop(\n    task=\"Take a screenshot of the desktop\",\n    api_key=\"sk-ant-...\",\n    loop_config=LoopConfig(\n        model=\"claude-sonnet-4-6\",\n        max_iterations=10,\n    ),\n    display_config=DisplayConfig(width=1920, height=1080),\n)\n\nprint(f\"Done in {result.iterations} iterations\")\nprint(result.final_text)\n```\n\n## API Versions\n\n| Model | Tool Version | Beta Flag |\n|-------|-------------|-----------|\n| Opus 4.6, Sonnet 4.6, O"},{"path":"_meta.json","content":"{\n  \"ownerId\": \"kn7d107jg9jv602h9ytsegydq184a42s\",\n  \"slug\": \"nm-phantom-computer-control\",\n  \"version\": \"1.9.19\",\n  \"publishedAt\": 1787750390470\n}"},{"path":"skill-card.md","content":"## Description:\n\nAutomates desktop GUI workflows via computer use API with screenshot capture.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[athola](https://clawhub.ai/user/athola)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and engineers use this skill to automate and verify GUI workflows that lack CLI or API alternatives, including visual web-app testing, form filling, and desktop app navigation.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: The skill can capture screenshots and synthesize keyboard and mouse input in other desktop applications.\n\nMitigation: Run it in a sandbox, VM, Docker, or controlled desktop environment, and close private windows before use.\n\nRisk: GUI actions can affect sensitive accounts or trigger irreversible real-world changes.\n\nMitigation: Avoid banking and sensitive services, and require explicit human confirmation before consequential actions.\n\n## Reference(s):\n\n- [ClawHub skill page](https://clawhub.ai/athola/skills/nm-phantom-computer-control)\n- [OpenClaw homepage](https://github.com/athola/claude-night-market/tree/master/plugins/phantom)\n\n## Skill Output:\n\n**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance]\n\n**Output Format:** [Markdown guidance with shell command and Python code examples]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May require desktop screenshot and input access; use iteration caps for cost and runaway-action control.]\n\n## Skill Version(s):\n\n1.9.19 (source: ClawHub release evidence)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":"Automates desktop GUI workflows via computer use API with screenshot capture Skill: computer-control Owner: athola Summary: Automates desktop GUI workflows via computer use API with screenshot capture Tags: latest:1.9.19 Version history: v1.9.19 | 2026-08-26T13:19:50.470Z | user Release v1.9.19 v1.9.17 | 2026-07-30T05:39:58.522Z | user Release v1.9.17 v1.9.16 | 2026-07-14T19:56:46.419Z | user Release v1.9.16 v1.9.14 | 2026-06-30T18:04:54.047Z | user Release v1.9.14 v1.9.13 | 2026-06-27T16:22:","editorialQuality":{"score":100,"threshold":65,"status":"ready","wordCount":919,"uniquenessScore":58,"reasons":[]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":"No screenshots, media assets, or demo links are available."},"primaryImageUrl":null,"mediaAssetCount":0,"assets":[],"demoUrl":null},"ownerResources":{"evidence":{"source":"unclaimed","verified":false,"confidence":"low","updatedAt":"2026-10-10T05:37:34.766Z","emptyReason":"This page has not been claimed by the agent owner."},"hasCustomPage":false,"customPageUpdatedAt":null,"customLinks":[],"structuredLinks":{"docsUrl":null,"demoUrl":null,"supportUrl":null,"pricingUrl":null,"statusUrl":null},"customPage":null},"relatedAgents":{"evidence":{"source":"protocol-neighbors","verified":false,"confidence":"medium","updatedAt":"2026-10-10T10:43:32.872Z","emptyReason":null},"items":[{"id":"8ebccd8e-3863-4187-8355-c3f14e1f9edf","entityType":"agent","canonicalPath":"/agent/iofficeai-aionui","slug":"iofficeai-aionui","name":"AionUi","description":"Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!","url":"https://github.com/iOfficeAI/AionUi","homepage":"https://www.aionui.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-10-09T19:11:12.944Z","createdAt":"2026-02-25T03:38:16.584Z","downloads":null},{"id":"b917f68a-ebff-438e-84f8-3f4b2494c0bc","entityType":"agent","canonicalPath":"/agent/activepieces-activepieces","slug":"activepieces-activepieces","name":"activepieces","description":"AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents","url":"https://github.com/activepieces/activepieces","homepage":"https://www.activepieces.com","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-15T02:22:12.426Z","createdAt":"2026-02-25T03:38:12.412Z","downloads":null},{"id":"5cb26759-3a39-483f-94cf-276a98c13bb8","entityType":"agent","canonicalPath":"/agent/cherryhq-cherry-studio","slug":"cherryhq-cherry-studio","name":"cherry-studio","description":"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs","url":"https://github.com/CherryHQ/cherry-studio","homepage":"https://cherry-ai.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-11T14:38:40.986Z","createdAt":"2026-02-25T03:38:19.379Z","downloads":null},{"id":"6f6582d0-5d76-4f0f-b81d-86520247950b","entityType":"agent","canonicalPath":"/agent/copilotkit-copilotkit","slug":"copilotkit-copilotkit","name":"CopilotKit","description":"The Frontend for Agents & Generative UI. React + Angular","url":"https://github.com/CopilotKit/CopilotKit","homepage":"https://docs.copilotkit.ai","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-03-25T09:50:57.846Z","createdAt":"2026-02-25T03:39:14.617Z","downloads":null}],"links":{"hub":"/agent","source":"/agent/source/clawhub","protocols":[{"label":"OpenClaw","href":"/agent/protocol/openclew"}]}}}