Server providing Argo CD integration for MCP clients.
Brickproof ADB 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.
What it does
Maintain your own Android devices and TV boxes over ADB with dry runs, guards and rollback
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user brickproof-adb-mcp '--' npx -y brickproof-adb
# 2 — ask your agent for something these tools do
# adb_devices, adb_connect, adb_pair and 15 more
Brickproof ADB costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
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 Brickproof ADB 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.
This is a public server: you run it yourself and this marketplace is not in the path. You need Node.js — the current LTS (24) is the safe choice; npx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user brickproof-adb-mcp '--' npx -y brickproof-adbReconnect, or start a new session, and the tools appear in the model’s tool list.
Nothing is required, but it can take BRICKPROOF_ADB (Path of the adb executable; optional, default adb on PATH), BRICKPROOF_FILE_ROOTS (Host directories adb_push, adb_pull and adb_install may use, separated by ; on Windows and : elsewhere; optional), BRICKPROOF_CONFIG (Path of the JSON config file; optional) if you want to set them.
One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.
18 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
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.
List connected Android devices with serial, state and description (adb devices -l), and remembered addresses that are not connected now. Start here: the other tools take serial from this list. If the device the user means is not in the list, call again with discover=true: it searches the local…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| discover | boolean | no | Search the local network (about 10 s) for devices with network or Wireless debugging on; returns candidates and the next step for each. Default false. |
Connect to a device over network ADB. host is ip or ip:port (default port 5555). An Android 11+ phone with Wireless debugging is paired once with adb_pair first; after that adb_connect is enough. The reverse is adb_disconnect.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host | string | yes | Device address: ip (port 5555 implied) or ip:port, e.g. 192.0.2.50 or 192.0.2.50:5555. |
Pair with an Android 11+ device over Wireless Debugging, once per device and computer; afterwards adb_connect alone reconnects. On the device: Developer options -> Wireless debugging -> "Pair device with pairing code" shows ip:port and a 6-digit code. The pairing port is random and differs from…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host | string | yes | ip:port from the pairing dialog (random port, NOT 5555) |
| code | string | yes | 6-digit pairing code from the dialog |
Disconnect a network ADB device, the reverse of adb_connect. Omit host to disconnect every network device. adb_disconnect then adb_connect recovers a device that adb_devices lists as offline. With host the address is also forgotten and not reconnected at start.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host | string | no | ip:port as listed by adb_devices. Omit to disconnect all network devices. |
Run a shell command on the device (settings, pm, am, dumpsys, getprop, wm, cmd, svc...) and return stdout. It bypasses every guard rail, so it works only while the server option allow_shell is on. For package changes use adb_app instead: dry runs, a protected set, a network guard and a rollback…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| command | string | yes | Command line for the device shell, e.g. "getprop ro.build.version.release". Logged verbatim with log_requests: type secrets with adb_text instead. |
| serial | no | — | |
| timeout_sec | number | no | Time limit in seconds, default 30, at most 120. Past it the call fails with "timed out". |
Take a screenshot of the device screen. Returns a JPEG image, then a text line with the logical screen size, the image size and the scale between them, e.g. "screen 1080x2340 -> image 473x1024 · scale 2.283/2.285". Pass coords="screenshot" to adb_tap/adb_swipe to tap coordinates read off that…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| serial | no | — | |
| quality | number | no | JPEG quality 30-95, default 70 |
| max_px | number | no | Max dimension 320-1920, default 1024 |
Dump the current UI hierarchy (uiautomator) as a compact list of interactive/labeled elements with tap coordinates @(x,y). Password fields are marked [PASSWORD] and listed even with no visible text. Use together with adb_tap. Prefer it over adb_screenshot for labels and tap targets; use…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| serial | no | — |
Tap at (x, y). coords="screen" (default): coordinates from adb_ui_dump, the logical space input tap uses. coords="screenshot": coordinates off the last adb_screenshot for the SAME serial, rescaled; missing or older than 120s is refused. Rotation between screenshot and tap is not checked.…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| x | number | yes | X in pixels, in the space chosen by coords. |
| y | number | yes | Y in pixels, same space. |
| serial | no | — | |
| coords | string | no | Coordinate space, default "screen". "screenshot" = pixels read off the last adb_screenshot image for this serial, rescaled automatically; refused if that screenshot is missing or older than 120s. |
| verify | string | no | Default "none". "activity" compares the resumed activity before/after. "ui" also hashes the UI dump (~2-4s); a clock or timer ticking on its own can make it report "changed". "unchanged" is reliable. |
Swipe from (x1,y1) to (x2,y2) over duration_ms (default 300). The same point at both ends with duration_ms >= 800 is a long-press. coords and verify work as in adb_tap and both endpoints share one coordinate space; with verify="ui", an "unchanged" result on a swipe usually means the end of a…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| x1 | number | yes | Start X in pixels, in the space chosen by coords. |
| y1 | number | yes | Start Y in pixels. |
| x2 | number | yes | End X in pixels. |
| y2 | number | yes | End Y in pixels. |
| duration_ms | number | no | Duration in ms, default 300. |
| serial | no | — | |
| coords | string | no | Coordinate space for BOTH endpoints, default "screen". See adb_tap for the "screenshot" semantics and its 120s TTL. |
| verify | string | no | See adb_tap. For a swipe, "ui" reporting "unchanged" usually means you hit the end of a scrollable list rather than that nothing happened. |
Type text into the focused input field. Focus the field first; on a TV box, DPAD_CENTER on the field opens the keyboard. For single keys (ENTER, BACK) use adb_key. ASCII goes through input text; non-ASCII (Cyrillic, emoji, CJK) is sent via the ADBKeyBoard IME, which must be installed on the device…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| text | string | yes | Text to type into the focused field. Only its length appears in the response and the log. |
| serial | no | — |
Send a keyevent. Accepts names (HOME, BACK, ENTER, DPAD_UP, POWER, VOLUME_UP, MENU, TAB...) or numeric codes. On TV boxes DPAD_UP/DOWN/LEFT/RIGHT move the focus and DPAD_CENTER presses it. verify="activity"|"ui" reports whether the key changed anything; see adb_tap for the same parameter. For text…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| key | string | yes | Key name (HOME, BACK, ENTER, DPAD_CENTER, WAKEUP...; the KEYCODE_ prefix is optional) or a numeric keycode. |
| serial | no | — | |
| verify | string | no | Default "none". See adb_tap. |
Find an on-screen element by text, resource-id or content-desc and activate it.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| text | string | no | Visible label to match, checked against both the text attribute and content-desc (substring by default, case-insensitive). Some TV launchers put every label in content-desc and leave text empty. |
| resource_id | string | no | resource-id; the part after "/" is enough |
| desc | string | no | content-desc to match |
| exact | boolean | no | Require an exact match instead of substring. Default false. |
| index | number | no | Which match to use when several elements match. Without it, several matches are an error listing the candidates. |
| max_steps | number | no | Max DPAD steps on leanback devices, default 12, max 40. Each step re-dumps the UI (~1.5-2s) and the walk also stops after an internal ~25s budget, returning a report instead of timing out. |
| serial | no | — |
Installs only with dry_run=false; by default it reports what would be installed and installs nothing.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| apk_path | string/array | no | Single APK path, array of split-APK paths, or one .apks/.xapk/.apkm bundle. Without it: a list of the installable files in file_roots, newest first; nothing is installed. |
| dir | string | no | Only without apk_path: the list shows the top level and one level of subfolders; for a deeper folder pass it here, listed the same way. Ignored when apk_path is given. |
| abi | string | no | Bundle only: force an ABI instead of the device default (e.g. armeabi-v7a) |
| density | number | no | Bundle only: force a screen density in dpi instead of the device value |
| locales | string/array | no | Bundle only: extra language splits to include alongside the device language (e.g. ["ru"] on an en-US device) |
| dry_run | boolean | no | Default true: report the files, the command and the device (for a bundle, also the chosen splits) and install nothing. Only false installs. |
| serial | no | — |
Uninstall an app by package name, at once and without a backup. keep_data=true keeps app data (-k). The reverse is adb_install. For system packages, reversible bulk operations and backups, use adb_app.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| package | string | yes | Package name, e.g. com.example.app (adb_app action=list lists them). |
| keep_data | boolean | no | Keep the app data and cache (-k). Default false. |
| serial | no | — |
Copy a file from the server's file roots (file_roots) to the device, overwriting a file of the same name there. To install an APK use adb_install instead. The reverse is adb_pull.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| host_path | string | yes | Absolute path of the file on the server, inside file_roots. |
| device_path | string | yes | Destination on the device, e.g. /sdcard/Download/file.apk or /data/local/tmp/. |
| serial | no | — |
Copy a file from the device into the server's file roots (file_roots), overwriting a file of the same name and creating missing directories. The reverse is adb_push.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| device_path | string | yes | File on the device, e.g. /sdcard/Download/report.pdf. |
| host_path | string | yes | Absolute destination path on the server, inside file_roots. |
| serial | no | — |
Read recent logcat lines without waiting for new ones, e.g. to see why an app crashed; unlike adb_shell it works with allow_shell off. filter: either a logcat filterspec like "ActivityManager:I *:S" (contains ":" or "*"; *:S is auto-appended if missing so unmatched tags are silenced) or a plain…
| Parameter | Type | Required | Notes |
|---|---|---|---|
| filter | string | no | A filterspec such as "ActivityManager:I *:S" (has ":" or "*"), or a plain substring matched case-insensitively. Omit for the last lines unfiltered. |
| lines | number | no | How many lines to return, default 200, at most 2000. |
| serial | no | — |
Package operations with guard rails: inspect, launch, stop, clear, disable, uninstall, back up and restore packages. Use it rather than adb_shell or adb_uninstall for package changes.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| action | string | yes | list/info/protected/state only read. launch/stop/clear act at once (clear defaults to dry_run), stop/clear under the network guard. disable/uninstall/enable/restore default to dry_run. |
| packages | string/array | no | Package name or array of package names. For action=launch with uri/intent_action, at most one package is accepted (used as -p to narrow the resolver); it is otherwise optional there. |
| uri | string | no | action=launch only: deep-link / intent data URI (am start -d). Its presence, together with intent_action, selects the intent-based launch instead of resolving the package's main activity. |
| intent_action | string | no | action=launch only: intent action, default android.intent.action.VIEW when uri is set. Must match [A-Za-z0-9_.]+. See the CALL refusal note above. |
| extras | object/string | no | action=launch only: intent extras. string -> --es, boolean -> --ez, int32 integer -> --ei, larger integer -> --el; anything else is refused by key name. A JSON-stringified object is accepted too. |
| serial | no | — | |
| filter | string | no | action=list only, default user |
| q | string | no | action=list only: case-insensitive substring filter |
| scope | string | no | action=backup only: what to back up when packages is omitted (default user) |
| mode | string | no | How to remove: disable (default) is undone by action=enable or restore; uninstall needs allow_uninstall. |
| dry_run | boolean | no | Default true for state-changing actions. Set false to actually apply. |
| batch_size | number | no | Packages per batch between canary checks, default 5, max 25 |
| canary | boolean | no | Check the account count and the launcher between batches, default true. Turning it off removes the only automatic check for a latent breakage. |
| backup | boolean | no | Pull APKs before removing. Default true for uninstall, false for disable. |
| force | boolean | no | Proceed even if the APK backup failed. Off by default. It does not affect the network guard; see force_network. |
| force_network | boolean | no | Override the network guard for disable, uninstall, stop and clear: act on a package that is currently serving an off-device client. It does NOT override the protected set. Off by default. |
| store | string | no | Where snapshots and APK backups live: by default the snapshot store (setting store). Another path must be inside file_roots. |
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under MIT
- 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 Brickproof ADB or to call it, at any volume.
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.
| 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 |
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.
No tools were enumerated, so prompt-injection, capability and toxic-flow analysis had no tool surface to inspect. To grade a package’s real runtime tools, 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 |
|---|---|---|
| No per-tool rows for this release — see the note below. | ||
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
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
Always latest — the install command names no version, so it installs the newest release when it runs; the current release on file is 1.0.0. Pin a release below and the command asks for exactly that one.
No release note was published with this version.
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.
Any signed-in account that does not publish Brickproof ADB can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
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.