Control a real Chrome browser to complete any task: fill forms, extract data, book flights.
Nothing to run from here
Moonlight Desktop is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
93/100
trust score
36
tools declared
local
runs on your machine
—
calls last 7d
What it does
Visual Windows desktop control over Moonlight/GameStream for MCP clients and coding agents
Quickstart
There is no one-line install for it here: it is set up the way its own source describes.
Publisher
Installation
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Moonlight Desktop as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
There is no one-line install for it here: it is set up the way its own source describes.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Read from its source code. A server that builds its tool list when it starts can list more tools than this.
Reports this MCP's installed version, local platform, native Moonlight bridge readiness, and canonical public setup and usage documentation. It sends no host request and does not change installations, profiles, or sessions.
No parameter schema on file.
Use this first when a task expects a paired remote Desktop. It reports configured remote targets, active Desktop sessions, every locally owned Moonlight transport that blocks another Desktop session, and a pending local launch preparation, without contacting a host or sending input. Select this…
No parameter schema on file.
Read the public GameStream serverinfo response. This never launches an app, pairs, or sends input.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host | string | yes | Host name or address, optionally with a GameStream HTTP port. |
Starts standard Apollo/GameStream pairing. Returns a four-digit PIN and the Apollo Web UI URL where the invoker can approve it. Generates a new local client identity; no existing Moonlight profile is modified.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host | string | yes | Host name or address, optionally with a GameStream HTTP port. |
| device_name | string | no | Friendly name shown in Apollo. Defaults to Moonlight MCP. |
Checks whether an Apollo PIN approval completed the current pairing handshake.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| pairing_id | string | yes | — |
Stops this MCP process from waiting for an Apollo PIN approval. It does not modify existing paired Apollo clients.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| pairing_id | string | yes | — |
Lists local Moonlight MCP profiles without returning client certificates or private keys. Use a returned id as profile_id for profile, wake, preflight, and session tools.
No parameter schema on file.
Stores the paired host's MAC address, IPv4 broadcast address, and UDP port locally. It does not send a magic packet. This enables host_wake and automatic wake attempts before session preflight/start when Apollo is offline.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | ID returned by profiles_list. |
| mac_address | string | yes | Host network-adapter MAC address, for example 00:d8:61:50:bf:75. |
| broadcast_address | string | no | IPv4 subnet broadcast address. Use the host LAN broadcast when global broadcast is filtered. |
| port | integer | no | UDP destination port for the magic packet. |
Sends Wake-on-LAN UDP magic packets from this MCP invoker using the selected profile's stored MAC and broadcast settings. It does not connect to Apollo, launch an app, or send desktop input. After waking, call session_preflight or session_start; those tools wait for Apollo when wake_if_offline is…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | ID returned by profiles_list. |
| attempts | integer | no | How many magic packets to send. |
| interval_ms | integer | no | Delay between magic packets. |
Lists apps through the selected paired MCP identity using mutual TLS. If Apollo has not granted List Apps, returns the exact permission request and Apollo Web UI URL. This never launches an app.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | — |
Reads the selected identity's authenticated Apollo permissions, decodes every granted and missing permission, and preflights visual computer use. This never launches an app or sends input.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | — |
Checks whether Apollo granted exactly the permissions required for an intended future session. Use mode computer_use for a normal Desktop session. If Apollo is offline and Wake-on-LAN is configured, wake_if_offline sends magic packets and waits for it to respond; otherwise this tool only reads…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | ID returned by profiles_list. |
| mode | string | no | Planned capability. Use computer_use for session_start or provider_app_start. |
| wake_if_offline | boolean | no | When configured, send Wake-on-LAN magic packets before reporting Apollo unavailable. |
| wake_timeout_ms | integer | no | Maximum time to wait for Apollo after an automatic wake. |
Launches Apollo's Desktop app and establishes a local Moonlight video/input session. This is the default and normal entry point for computer use. The full_hd stream profile defaults to 1920x1080 at 30fps and 20 Mbps; select low_bandwidth for 1280x720 at 30fps and 8 Mbps when the connection is…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | ID returned by profiles_list. |
| stream_profile | string | no | full_hd uses 1920x1080 at 30fps and 20 Mbps. low_bandwidth uses 1280x720 at 30fps and 8 Mbps for constrained connections. |
| width | integer | no | Optional override for the selected Desktop stream profile width. |
| height | integer | no | Optional override for the selected Desktop stream profile height. |
| fps | integer | no | Optional override for the selected Desktop stream profile frame rate. |
| bitrate_kbps | integer | no | Optional override for the selected Desktop stream profile bitrate. |
| wake_if_offline | boolean | no | When configured, send Wake-on-LAN magic packets before reporting Apollo unavailable. |
| wake_timeout_ms | integer | no | Maximum time to wait for Apollo after an automatic wake. |
Explicit provider-app path. Use this only when the user expressly requests an Apollo/Moonlight registered app rather than visual Desktop computer use. If configured, it can Wake-on-LAN before launch. The result's sessionId is required by capture, input, status, and stop tools. It never takes over…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| profile_id | string | yes | ID returned by profiles_list. |
| app_id | integer | yes | Explicit Apollo provider app ID. |
| app_name | string | yes | Name of the explicitly requested provider app. |
| width | integer | no | — |
| height | integer | no | — |
| fps | integer | no | — |
| bitrate_kbps | integer | no | — |
| wake_if_offline | boolean | no | When configured, send Wake-on-LAN magic packets before reporting Apollo unavailable. |
| wake_timeout_ms | integer | no | Maximum time to wait for Apollo after an automatic wake. |
Returns local Moonlight transport state and the timestamp of its newest video frame. This never sends remote input.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
Returns the latest decoded Moonlight frame as a PNG image. A cached frame is used only when it is at most max_age_ms old; otherwise the bridge requests a new video frame and waits for it. The 500 ms default is intended for visual computer use.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| max_age_ms | integer | no | — |
Returns a PNG crop in the active session's capture-space coordinates. Supply a snapshot ID from screen_capture to crop the exact observed image, or omit it to capture a current frame first.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| snapshot_id | string | no | — |
| x | integer | yes | — |
| y | integer | yes | — |
| width | integer | yes | — |
| height | integer | yes | — |
| max_age_ms | integer | no | — |
Compares a retained baseline screenshot with another retained or current screenshot. Returns the proportion and bounding box of material pixel changes in capture-space coordinates.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| baseline_snapshot_id | string | yes | — |
| comparison_snapshot_id | string | no | — |
| pixel_threshold | integer | no | — |
| max_age_ms | integer | no | — |
Waits for the active Desktop stream to differ materially from a retained screenshot, then returns the resulting frame and changed region. Use after actions that open a window, load content, or change a setting. It ignores tiny cursor/compression changes by default.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| baseline_snapshot_id | string | no | — |
| timeout_ms | integer | no | — |
| min_changed_fraction | number | no | — |
| pixel_threshold | integer | no | — |
| region_x | integer | no | — |
| region_y | integer | no | — |
| region_width | integer | no | — |
| region_height | integer | no | — |
Stores a cropped region from a retained screenshot as a session-local visual template. Use it with screen_find_template to locate the same icon, control, or image in later Desktop frames.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| snapshot_id | string | yes | — |
| x | integer | yes | — |
| y | integer | yes | — |
| width | integer | yes | — |
| height | integer | yes | — |
Finds a session-local visual template in a current Desktop screenshot and returns its capture-space bounding box and similarity score. The result is deterministic pixel matching; use screen_capture for semantic interpretation.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| template_id | string | yes | — |
| min_score | number | no | — |
| search_x | integer | no | — |
| search_y | integer | no | — |
| search_width | integer | no | — |
| search_height | integer | no | — |
| max_age_ms | integer | no | — |
Runs local OCR on a current or retained Desktop screenshot and returns recognized text plus capture-space line and word boxes. The first use of a language may download its OCR data on the MCP invoker machine; no Windows software is installed.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| snapshot_id | string | no | — |
| language | string | no | Tesseract code, for example eng, por, or eng+por. |
| min_confidence | number | no | — |
| max_results | integer | no | — |
| region_x | integer | no | — |
| region_y | integer | no | — |
| region_width | integer | no | — |
| region_height | integer | no | — |
| max_age_ms | integer | no | — |
Runs local OCR and finds matching visible text. Returns capture-space boxes suitable for session_mouse_click_at. Searches full recognized lines first, then individual words when no line matches.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| query | string | yes | — |
| snapshot_id | string | no | — |
| language | string | no | — |
| min_confidence | number | no | — |
| max_results | integer | no | — |
| region_x | integer | no | — |
| region_y | integer | no | — |
| region_width | integer | no | — |
| region_height | integer | no | — |
| max_age_ms | integer | no | — |
Polls local OCR until requested text is visibly present or a timeout expires. Use for menus, completion notices, and dialogs; successful matches include capture-space boxes for the next visual action.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| query | string | yes | — |
| language | string | no | — |
| min_confidence | number | no | — |
| timeout_ms | integer | no | — |
| poll_interval_ms | integer | no | — |
| region_x | integer | no | — |
| region_y | integer | no | — |
| region_width | integer | no | — |
| region_height | integer | no | — |
Queues an absolute mouse position over the active encrypted Moonlight input channel. Coordinates are in the current screen_capture pixel space.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| x | integer | yes | — |
| y | integer | yes | — |
Queues a complete mouse click over the active encrypted Moonlight input channel, holding the button briefly so the Windows desktop can recognise it reliably.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| button | string | no | — |
| count | integer | no | — |
Moves to a capture-space coordinate, lets the remote desktop observe the position, then sends a complete click. Prefer this atomic visual-computer-use action over separate move and click calls when clicking a target found in screen_capture.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| x | integer | yes | — |
| y | integer | yes | — |
| button | string | no | — |
| count | integer | no | — |
Moves to a starting capture-space coordinate, holds the selected mouse button, sends interpolated absolute positions over the encrypted Moonlight input channel, and releases the button. Use screen_capture before and after the drag.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| start_x | integer | yes | — |
| start_y | integer | yes | — |
| end_x | integer | yes | — |
| end_y | integer | yes | — |
| button | string | no | — |
| duration_ms | integer | no | — |
Queues a high-resolution vertical wheel event over the encrypted Moonlight input channel. Amount is in 1/120-wheel-click units; positive is up and negative is down.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| amount | integer | yes | — |
Queues UTF-8 text through Moonlight's encrypted text-input event. Use screen_capture to verify the visible result.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| text | string | yes | — |
Queues a Windows virtual-key event through Moonlight. Use action down/up for shortcuts; modifiers is a Moonlight bitmask: Shift 1, Ctrl 2, Alt 4, Meta 8, Extended 16.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| virtual_key | integer | yes | — |
| action | string | yes | — |
| modifiers | integer | no | — |
Sends one complete Windows virtual-key press through Moonlight. Use session_shortcut for multi-key chords such as Ctrl+Z, and session_key only when an individual key must remain held.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| virtual_key | integer | yes | — |
| modifiers | integer | no | — |
Presses and releases a named Windows key such as ENTER, ESCAPE, TAB, SPACE, F5, LEFT, or WINDOWS. Prefer this over raw virtual-key codes for ordinary visual computer use.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| key | string | yes | — |
Sends a named key chord as real key-down events followed by reverse key-up events. Examples: key Z with modifiers [CTRL] for undo, key F4 with [ALT] to close a window, or key L with [WINDOWS].
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| key | string | yes | — |
| modifiers | array | no | — |
Sends a chord as real key-down events in order and key-up events in reverse order. For example, [17, 90] is Ctrl+Z. This is more reliable for desktop shortcuts than encoding a modifier bit on one key event.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
| virtual_keys | array | yes | Windows virtual-key codes in press order. |
Stops the local Moonlight transport and asks Apollo to cancel the Desktop app launched by this MCP session. It does not affect unrelated existing sessions, which this prototype never takes over.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session_id | string | yes | — |
Free
$0 forever
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source is public, though no licence is declared for it
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install Moonlight Desktop or to call it, at any volume.
Call allowance · this installenforced at the gateway
Monthly allowance, in calls
—
—
Alert at 80% · then the tool returns a normal MCP error
Used this month: —
ATrust score 93 / 100
Release 0.10.0 · scanned 24 Oct 2026
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
Threat score100
Client score93
Findings8 threat · 0 capability
Capabilityhigh
Coveragesource
Verificationrepo
Surface digestea2e776cf6a0375a538e4dd4b9758bc59fedcd24d4b34ae3c23195a96fdafe9e
Score vectorthreat 100 → client 93
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | high | capability blast radius (high) — client exposure if the model is manipulated | −6 |
| verification-discount | repo | publisher verification (public source) — no provenance, but the source is public and inspectable | −1 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
score = clamp(round(threatScore − E_cap − E_ver − E_cov), 0, 100). No model, no opinion, no network call at scoring time: the same methodology version against the same bytes always produces this number, and every line above is a field of the report rather than a summary of it.
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
Tool surfaceread
Implementation sourceread
Package metadataread
Live transportnot available
Tools were statically extracted from the published source (36 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. high is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| runtime_status | no tags | |
| desktop_channel_status | no tags | |
| host_status | no tags | |
| pairing_begin | state-write | state-write: declares readOnlyHint: false |
| pairing_status | no tags | |
| pairing_cancel | state-write | state-write: declares readOnlyHint: false |
| profiles_list | no tags | |
| profile_wol_configure | state-write | state-write: declares readOnlyHint: false |
| host_wake | state-write | state-write: declares readOnlyHint: false |
| apps_list | no tags | |
| profile_status | no tags | |
| session_preflight | state-write | state-write: declares readOnlyHint: false |
| session_start | state-write | state-write: declares readOnlyHint: false |
| provider_app_start | state-write | state-write: declares readOnlyHint: false |
| session_status | no tags | |
| screen_capture | no tags | |
| screen_crop | no tags | |
| screen_diff | no tags | |
| session_wait_for_change | no tags | |
| screen_template_create | no tags | |
| screen_find_template | no tags | |
| screen_ocr | untrusted-input | untrusted-input: keyword "download" |
| screen_find_text | no tags | |
| screen_wait_for_text | no tags | |
| session_mouse_move | state-write | state-write: declares readOnlyHint: false |
| session_mouse_click | state-write | state-write: declares readOnlyHint: false |
| session_mouse_click_at | state-write | state-write: declares readOnlyHint: false |
| session_mouse_drag | state-write | state-write: declares readOnlyHint: false |
| session_scroll | state-write | state-write: declares readOnlyHint: false |
| session_type | state-write | state-write: declares readOnlyHint: false |
| session_key | state-write | state-write: declares readOnlyHint: false |
| session_key_press | state-write | state-write: declares readOnlyHint: false |
| session_key_press_named | state-write | state-write: declares readOnlyHint: false |
| session_hotkey | state-write | state-write: declares readOnlyHint: false |
| session_shortcut | state-write | state-write: declares readOnlyHint: false |
| session_stop | state-write | state-write: declares readOnlyHint: false |
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.
Supply chain and provenance
Integrityfirst-seen
Surface digestea2e776cf6a0375a
Scanned2026-10-24
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
It installs from its repository rather than from a versioned package, so the command cannot be pinned to a release. The history below is for reading; the current release on file is 0.10.0.
0.10.0
23 Oct 2026
latestNo release note was published with this version.
0.9.1
undated
- The version this listing was on when mcprush began following its releases. The registry was not asked when it shipped, so this row carries no date.
—
no reviews yet
Who can review
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Sign in to write one
Any signed-in account that does not publish Moonlight Desktop can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
No reviews yet
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.