dsh-cua - Windows Computer Use
PYPI · DSH-CUA
Windows computer-use MCP server for dsh (DeepSeek Harness), client-neutral, 19 tools.
Available components
Last scanned (11 hours ago)
Scores higher than 65% of PyPI servers
Category Browser Automation
How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →
Supply Chain Security41
- Malware scan not yet available for this package.Unverified
- CVE check failed: a known advisory affects this package; its severity couldn't be graded. See how to fix → View diagnostics → Fail
- Runs setuptools.build_meta at install time, a recognised build step with no custom scripting around it. View diagnostics → Pass
- 1 of 36 dependencies flagged as unhealthy. View diagnostics → Partial
Provenance & Transparency97
- Source repository is publicly reachable at the declared URL. View diagnostics → Pass
- Cryptographically verified build provenance (signed, bound to Hutusion/dsh-cua). View diagnostics → Pass
- Clear OSI-approved license (MIT).Pass
- Actively maintained (last published 0 days ago).Pass
- Disclosure check failed: no security disclosure policy was found in the source repository. See how to fix → Fail
Schema Quality & AI Usability68
- AI-judged instruction clarity (good).Pass
- Context-footprint check failed: tool/resource definitions use about 2479 tokens (~130/item across 19 items; 19 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management100
- No destabilizing schema changes in the last 30 days.Pass
Tool Coverage71
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 0% of tool parameters carry a description.Fail
- Structured output schemas are declared (100% of tools); any adoption earns full credit.Pass
Tool Safety75
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- 0 of 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation; "tool_send_keys" implies "send" and declares no destructiveHint at all, which the MCP spec reads as destructive by default. See how to fix → Fail
- An AI judge read all 20 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.Pass
Checks we ran (15)
- Known CVEs
- Install scripts
- Dependency health
- Source repository
- Provenance
- License
- Maintenance
- Security disclosure
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Checks we could not complete (1)
These are gaps on our side, not findings about this package. We only credit what we can confirm, so a category made up entirely of them scores 0.
- Malware scan in Supply Chain SecurityMalware scan not yet available for this package.
Does not apply to this package (2)
- Endpoint SecurityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
- Transport & ReachabilityScored on hosted endpoints only. A package has no live endpoint of its own for us to probe.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 21 Nov 26 0
- Package version: 0.4.1 → 0.5.0 functional
- 20 Nov 26 −15
- Malware scan: pass → unverified ▼ security
- Stability: 0.97 → pass security
- 19 Nov 26 0
- Stability: pass → 0.97 functional
- 18 Nov 26 +15
- Malware scan: unverified → pass ▲ security
- 17 Nov 26 −15
- Malware scan: pass → unverified ▼ security
- 13 Nov 26 +15
- Malware scan: unverified → pass ▲ security
- Stability: 0.97 → pass security
- 12 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 93 to 97. That category is still filling its 30-day observation window: 28 days of observed history at the previous scan, 29 at this one. The score rises as the window fills, whether or not the server changes.
- 11 Nov 26 −16
- Malware scan: pass → unverified ▼ security
- Stability: pass → 0.93 functional
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Analysed pypi/dsh-cua@0.5.0
Provenance Verified
A signed build attestation was found and verified, binding this exact artifact to the source repository it claims to come from.
| Result | Verified |
|---|---|
| Ecosystem | pypi |
| Reason | Verified |
| Discovered via | Registry attestation endpoint |
| Source repo | Hutusion/dsh-cua |
| Certificate issuer | https://token.actions.githubusercontent.com |
| Certificate SAN | https://github.com/Hutusion/dsh-cua/.github/workflows/publish.yml@refs/tags/v0.5.0 |
| Rekor log index | 4030980034 |
| Predicate type | PyPI publish attestation https://docs.pypi.org/attestations/publish/v1 |
| Subject digest | sha256:db5c0b8bb8c792446dd08e4c412c92168e6ae3b2fc087b40e827aa01baee830b |
Background: How many MCP packages publish verified provenance →
Install scripts 1 script
| Hook | Tier | Command |
|---|---|---|
| build_backend | Known build tool | setuptools.build_meta |
Background: Why install scripts are a supply-chain risk →
Vulnerabilities 2 findings
| ID | CVE | Severity | Vector | Fix available |
|---|---|---|---|---|
| GHSA-hwfp-hg2m-9vr2 | CVE-2021-32559 | medium | CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H | yes |
| PYSEC-2021-112 | CVE-2021-32559 | none | yes |
Background: What a vulnerability scan can and cannot prove →
Dependencies 36 packages
| Packages resolved | 36 |
|---|---|
| Stale | 1 |
| Tree resolution | Complete |
Background: SBOMs and build attestations, explained →
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
tool_capture_window ~323
Capture a window screenshot and SAVE IT TO A FILE. Provide hwnd or title substring. Crops to the window's CLIENT area by default, so image pixel (0,0) is the same point click_at calls (0,0) — no frame offset to guess. Returns bounds = the screen rect (x,y,width,height) the image maps to, and per-axis `scale_x` and `scale_y`. Multiply image coordinates by the corresponding scale and then add the bounds origin to get screen coordinates. A DPI-virtualized target can have different image-pixel and screen-pixel dimensions even without resizing. Also reports dpi_verified; when false, image pixels and input coordinates cannot be safely paired. Defaults to JPEG q80 — a 1080p shot is ~6-10x smaller than PNG, which is what keeps a long session's request body under the provider's 32 MiB cap. The bytes are NOT inlined as base64 by default (that costs ~12,500 tokens per shot as text); read the saved_path with read_image instead, which attaches the same picture for ~50 tokens. Pass include_data_uri=true only when a caller genuinely cannot read a file. Read-only — no input injection, no focus change.
| Name | Type | Req | Description |
|---|---|---|---|
| client_only | boolean | – | – |
| hwnd | integer | – | – |
| image_format | string | – | – |
| include_data_uri | boolean | – | – |
| max_dim | integer | – | – |
| quality | integer | – | – |
| save_path | string | – | – |
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_click_at ~269
Click at client-area coordinates within a window. Coordinates relative to window content area (0,0 = top-left). Tries the ELEMENT path first (resolve the point via UI Automation and invoke that element) because it needs no focus and does not care about z-order; falls back to a gated raw cursor click before dispatch when the hit is an image/document surface or cannot be invoked. A sent Invoke is never followed automatically by a raw click. Check effect_verified and observe the window when the effect is unconfirmed. The result's `method` says which was used: 'ax_press' or 'raw_event'. MUTATING: serialized across agent sessions; the raw_event path ALSO YIELDS TO THE USER (waits for input-quiet, then refuses with user-active rather than fighting them for the cursor); the ax_press path injects no physical input. The raw path verifies the addressed window actually owns that screen point and returns success=false instead of clicking another window. Pass dry_run=true to check where the click would land without moving the cursor, clicking, or taking any gate.
| Name | Type | Req | Description |
|---|---|---|---|
| dry_run | boolean | – | – |
| hwnd | integer | – | – |
| prefer | string | – | – |
| title | string | – | – |
| x | integer | – | – |
| y | integer | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_clipboard_read ~30
Read the current clipboard text. Read-only. Non-text clipboard content reports empty text rather than failing.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
tool_clipboard_write ~58
Replace the clipboard contents with text. MUTATING (soft): serialized across agent sessions — and it DESTROYS whatever the user last copied, so do not use it casually; announce it when you do.
| Name | Type | Req | Description |
|---|---|---|---|
| text | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_coexistence_status ~88
Report the coexistence policy in force: whether the admission gate is enabled, the quiet/max-wait/mutex-timeout tunables, the pre-injection final-check window, named-mutex health, and how long ago the last physical input happened. Read-only — call it when an action was refused with user-active or arbiter-busy and you want the numbers behind the refusal.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
tool_cursor_position ~24
Current cursor position plus the screen size, both physical pixels. Read-only.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
tool_element_action ~169
Perform an action on a UI element by the `ref` from element_at_point — this is the focus-free path: a UI Automation pattern is delivered to the element, so the window is never raised and z-order never matters. MUTATING (soft): serialized across agent sessions; NO physical input is injected, so it can run while the user types — just do not operate the very window the user is working in. `action` must be one of the actions listed for that element (press, set_value, select, toggle, expand, collapse, scroll_into_view, focus); `set_value` also needs `text`. Refuses with the supported list when the element does not offer the action.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| ref | string | yes | – |
| text | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_element_action_at ~158
Act on an element printed by the latest skyshot of a window. Pass hwnd, shot_id and the element's printed index from the same tool_skyshot response. A missing or stale shot_id or an index omitted from the response is refused before acting. If refused, take a fresh full shot with disable_diff=true and use its shot_id and index. MUTATING (soft): serialized across agent sessions; no physical input injected. Actions: press, set_value, select, toggle, expand, collapse, scroll_into_view, focus. set_value also needs text.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| hwnd | integer | yes | – |
| index | integer | yes | – |
| shot_id | string | – | – |
| text | – | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_element_at_point ~108
Identify the UI element at a PHYSICAL screen point via Windows UI Automation. Returns its name, role, rect, and which actions it supports, plus an opaque `ref` for read_element. This is how you turn 'what I see in the screenshot' into something addressable without clicking: coordinates here are the same space as capture bounds (both physical pixels). Read-only — moves no cursor and changes no focus.
| Name | Type | Req | Description |
|---|---|---|---|
| x | integer | yes | – |
| y | integer | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_find_elements ~190
Find UI elements by role and/or name substring and return refs that can be acted on with element_action. This is how you reach an element that cannot be hit-tested — the point of naming it is that you no longer need a pixel. `role` is the lowercase role from skyshot (button, edit, document, list_item, ...). At least one filter is required: an unfiltered call would be a tree dump, so use skyshot for that. The result reports `minimized` — a minimized window does not always expose its full tree, so a zero here is not proof that nothing matches. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| automation_id | string | – | – |
| hwnd | integer | – | – |
| include_offscreen | boolean | – | – |
| max_results | integer | – | – |
| name_contains | string | – | – |
| role | string | – | – |
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_find_window ~32
Find a window by title substring and return its hwnd and geometry. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_get_window_rect ~47
Get detailed position and size info for a window: absolute position, dimensions, and client area size. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| hwnd | integer | – | – |
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_list_displays ~38
List monitors: index (1-based), primary flag, device name, and the monitor plus work-area rectangles in physical pixels. Read-only.
Input schema present but exposes no named parameters.
Structured output declared, but exposes no named fields.
No examples provided.
tool_list_windows ~47
List all visible top-level windows with title, hwnd, PID, position and size, sorted largest first. Read-only — safe while the user works.
| Name | Type | Req | Description |
|---|---|---|---|
| filter_title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_open_application ~68
Open a file, folder, or URI with the shell's default handler — like double-clicking it in Explorer: an .exe path, a document, or a URL. MUTATING (soft): serialized across agent sessions; the launched app may take focus.
| Name | Type | Req | Description |
|---|---|---|---|
| target | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_read_element ~68
Re-read a UI element by the `ref` returned from element_at_point. Use this to check whether an element is still there before acting on it; a ref goes stale when the UI is rebuilt, and that is reported rather than guessed. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| ref | string | yes | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_send_keys ~175
Send keyboard shortcut to a window. Example: keys=['ctrl','s']. Physical keys follow the active input method. A successful dispatch does not verify that text was committed. Use set_value for literal text when available. For Alt+key menu navigation, include 'alt' as first key (or set use_alt_key). MUTATING: serialized across agent sessions via the admission gate, and global hotkeys additionally YIELD TO THE USER — recent mouse/keyboard activity makes them wait, then refuse with reason user-active instead of interrupting; switch to element actions or retry later. Also refuses when the window is not the foreground window, because injected keys would go to the wrong window.
| Name | Type | Req | Description |
|---|---|---|---|
| hwnd | integer | – | – |
| keys | array | – | – |
| title | string | – | – |
| use_alt_key | boolean | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_skyshot ~258
Read a window's accessibility tree as compact text. Each element line is '{index} {indent}{role} {name}{Value: ...}{(state)}'. After the first shot, the result is a diff by default: '~' shows changed or renumbered elements, '+' shows additions, and removed indices refer to the previous shot. Unchanged lines are omitted. For tool_element_action_at, pass the returned hwnd and shot_id plus an index printed on an element line in this response. Taking another shot of the same window invalidates that pair. Pass disable_diff=true to see the full tree and act on an omitted element. The shot_key identifies the window, while shot_id identifies this observation. Pass include_offscreen=true to include scrolled-out or hidden elements. A minimized window may expose only part of its tree; restore the window and read again if needed. Element values can contain window content. Read-only: no input injection or focus change.
| Name | Type | Req | Description |
|---|---|---|---|
| disable_diff | boolean | – | – |
| hwnd | integer | – | – |
| include_offscreen | boolean | – | – |
| include_values | boolean | – | – |
| max_depth | integer | – | – |
| max_nodes | integer | – | – |
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
tool_type_text ~128
Post text to a window's text control without activating or restoring its window. MUTATING (soft): serialized across agent sessions. No physical input is injected. On PostMessage failure, stops at the first rejected character and returns how many were queued in typed. A partial post may already have edited the target. Inspect it before retrying. effect_verified reports whether readback confirmed the requested insertion. Some applications require their window to be active to process text.
| Name | Type | Req | Description |
|---|---|---|---|
| delay_ms | integer | – | – |
| hwnd | integer | – | – |
| text | string | – | – |
| title | string | – | – |
Structured output declared, but exposes no named fields.
No examples provided.
How do I install the dsh-cua - Windows Computer Use MCP server?
dsh-cua - Windows Computer Use runs locally as a PyPI package, launched with uvx dsh-cua. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
pypi · dsh-cua
claude mcp add hutusion-dsh-cua -- uvx dsh-cua
What is the dsh-cua - Windows Computer Use MCP server?
dsh-cua - Windows Computer Use is an MCP server listed in the public MCP registry as io.github.Hutusion/dsh-cua. Windows computer-use MCP server for dsh (DeepSeek Harness), client-neutral, 19 tools. This page covers its PyPI package (dsh-cua).
Is the dsh-cua - Windows Computer Use MCP server safe to use?
dsh-cua - Windows Computer Use scores 72 out of 100 on VerifyMCP. We recorded 2 known advisories against it as of 21 November 2026. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the dsh-cua - Windows Computer Use MCP server expose?
dsh-cua - Windows Computer Use exposes 19 tools: tool_list_windows, tool_find_window, tool_capture_window, tool_send_keys, tool_click_at, and 14 more. Their descriptions and schemas cost roughly 2,278 tokens of context every time the server is loaded.
Is the dsh-cua - Windows Computer Use MCP server still maintained?
dsh-cua - Windows Computer Use is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What licence is the dsh-cua - Windows Computer Use MCP server under?
dsh-cua - Windows Computer Use declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.
What did VerifyMCP test on the dsh-cua - Windows Computer Use MCP server?
VerifyMCP ran 15 checks on dsh-cua - Windows Computer Use, covering Known CVEs, Install scripts, Dependency health, Source repository and 11 more. One item could not be completed, a gap on our side rather than a finding about dsh-cua - Windows Computer Use: Malware scan. Endpoint Security and Transport & Reachability do not apply to a PyPI package, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.