cua-driver-f48465 — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited cua-driver-f48465 (Agent Skill) and scored it 92/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 2 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 2 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Use DeepChat's plugin-provided Computer Use tools as the only action surface for this skill. Do not ask the user to install cua-driver, configure an external server, or put anything on PATH for the bundled DeepChat plugin.
${OWNER_PLUGIN_ID}.${PLUGIN_ROOT}.${PROCESS_ARCH}.darwin/arm64, darwin/x64, win32/x64, win32/arm64,linux/x64.
linux/arm64.DeepChat.app/Contents/Helpers/DeepChat Computer Use.app; the plugin-local fallback is ${PLUGIN_ROOT}/runtime/darwin/${PROCESS_ARCH}/DeepChat Computer Use.app.
${PLUGIN_ROOT}/runtime/win32/${PROCESS_ARCH}/cua-driver.exe.${PLUGIN_ROOT}/runtime/linux/${PROCESS_ARCH}/cua-driver.list_apps. Match localized names, English names, romanized names, bundleidentifiers, executable names, and common abbreviations. Prefer stable identifiers when a result provides them.
launch_app. Use the returned pid when available.list_windows({ pid }) when the launch result lacks a usable window.get_window_state({ pid, window_id }). Use a visioncapture mode when visual evidence is needed.
click, right_click, double_click, drag, scroll,type_text, press_key, hotkey, set_value, page, or launch_app with URLs/files when supported by the platform.
playback progress, new panels, highlighted rows, or updated window content.
Element indices come from the latest get_window_state result for the same pid and window_id. Re-snapshot when an index is missing, stale, or from another window.
check_permissions for Accessibility and Screen Recording status. If a grant ismissing, ask the user to grant it to the detected DeepChat Computer Use.app helper opened by DeepChat.
list_apps, then calllaunch_app with a Windows name, path, launch_path, or aumid. Do not use macOS bundle ids on Windows. Use bring_to_front only when foreground interaction is necessary for the task.
unavailable. Use extra snapshots and report platform limits clearly when a tool cannot complete.
Many media, browser, and Electron apps expose a shallow accessibility tree while still showing actionable pixels.
Use this fallback order:
get_window_state({ pid, window_id }) when the first tree is sparse.page when DOM access identifies the target more reliably thanpixels.
get_window_state with vision capture for broad visual confirmation when window contents oractive overlays are unclear.
zoom({ pid, window_id, x1, y1, x2, y2 }) for small text or dense icons.Repeated zoom calls are a failure signal; return to the full-window snapshot or ask for clarification.
click({ pid, window_id, x, y }),or from the single zoom image with click({ pid, window_id, x, y, from_zoom: true }).
Ask the user only when visible candidates are ambiguous, the requested action is destructive, or the target is outside the current visible window.
launch_app.launch_app with the platform-supported file or URLarguments.
window_id.app is active enough for the platform to expose menu state reliably.
Use get_agent_cursor_state to inspect the cursor overlay. Use set_agent_cursor_enabled, set_agent_cursor_motion, or set_agent_cursor_style only when the user asks to show, hide, animate, or restyle the agent cursor.
Use start_recording, stop_recording, get_recording_state, and replay_trajectory for recording workflows. Use install_ffmpeg only with explicit user approval.
README.md: compact workflow reference.WEB_APPS.md: browser and webview patterns.RECORDING.md: recording and replay tool notes.TESTS.md: manual verification scenarios.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.