snipaya-mcp
This server acts as a bridge between an AI assistant and the Snipaya macOS app, enabling the assistant to interact with, view, and annotate screenshot canvases.
Content Analysis: Read and describe the current canvas contents, perform OCR on images, and pick specific pixel colors.
Annotation: Add diverse elements to a canvas, including arrows, shapes (rectangles, ellipses), text, callouts, spotlights, freehand strokes, and step numbers.
Image Manipulation: Crop, redacts sensitive data like emails and tokens (using blur, noise, or solid boxes), and manage tabs (add new, switch, close).
Workflow & Output: Export the finished canvas as a PNG/JPG, save projects, or send the canvas to various network destinations.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@snipaya-mcp blur the email on my canvas and add a red arrow pointing to it"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Connect any MCP client that speaks stdio to Snipaya, the image and screenshot annotator for macOS. Your assistant can read the canvas, add arrows, text, shapes and callouts, redact sensitive data, crop, and export, while you watch the result in Snipaya.
Snipaya has an MCP server built in, served over Streamable HTTP on 127.0.0.1. snipaya-mcp is a thin bridge for clients that only start local stdio servers: it forwards every MCP message between the client and Snipaya, unchanged.
Tools and prompts are listed even before Snipaya is open. Until Snipaya answers, the bridge lists them from a built-in catalog, and a tool call returns a message that says what to do.
Listed on the MCP Registry as io.github.krzysztofsurdy/snipaya-mcp, on Glama and in TensorBlock awesome-mcp-servers.
Requirements
Snipaya for macOS. It comes with a 14-day free trial.
Node.js 20 or newer (for the
npxsetup).
Related MCP server: @pixelvault-dev/local-mcp
Turn on the MCP server in Snipaya
The server is off by default.
Open Snipaya, then Settings > AI Assistants.
Turn on Allow AI assistants. The status changes to Running.
That is all the bridge needs. It reads the access token and port from the settings Snipaya saves on this Mac, so you do not have to copy them.
Setup
npx
npx -y snipaya-mcpIf Snipaya is not running but Allow AI assistants is on, the bridge opens Snipaya in the background and waits up to 10 seconds for its server.
To override what Snipaya saved, set these variables or flags. Flags win over variables, and variables win over the saved settings.
Variable | Flag | Default | |
|
| Token Snipaya saved | Access token from Settings > AI Assistants. |
|
|
| Server URL from Snipaya. Only |
| Set to |
Prefer the environment variable over --token: command-line flags are visible to other processes on the Mac.
JSON config
For clients configured with an mcpServers JSON file:
{
"mcpServers": {
"snipaya": {
"command": "npx",
"args": ["-y", "snipaya-mcp"],
"env": {
"npm_config_registry": "https://registry.npmjs.org/"
}
}
}
}For a custom setup, pass the values in env instead:
{
"mcpServers": {
"snipaya": {
"command": "npx",
"args": ["-y", "snipaya-mcp"],
"env": {
"npm_config_registry": "https://registry.npmjs.org/",
"SNIPAYA_MCP_TOKEN": "<access token>",
"SNIPAYA_MCP_URL": "<server url>"
}
}
}
}If your client supports Streamable HTTP servers directly, you do not need this bridge: use Copy MCP config in Snipaya instead.
MCP Bundle (.mcpb)
Download
snipaya-mcp-<version>.mcpbfrom the latest release.Open it with an MCP client that supports MCP Bundles, or drag it into the client's extensions settings.
Leave the token and URL empty to use what Snipaya saved, or paste the values from Settings > AI Assistants for a custom setup.
Security model
Local only. Snipaya's server listens on
127.0.0.1and accepts only loopbackHostheaders and requests without anOriginheader, so websites in your browser cannot reach it. The bridge refuses any URL whose host is not127.0.0.1orlocalhost, and never sends anOriginheader.Off by default. Nothing listens until you turn on Allow AI assistants.
Token. Every request carries
Authorization: Bearer <token>. Snipaya generates the token, and Regenerate in Settings revokes the old one. The bridge never writes the token to its logs.Saved settings. Without
SNIPAYA_MCP_TOKEN, the bridge reads the token and port from~/Library/Application Support/Snipaya/mcp.json, which Snipaya writes readable only by your macOS user. The bridge never writes to it.Opening Snipaya. Only on macOS and only when Allow AI assistants is on, the bridge first sends one request without the token to see whether Snipaya answers. If the connection is refused, it runs
open -g -b dev.snipaya.app(no shell) and retries. It does this at start, and again for a tool call while Snipaya is not reachable. Any other failure, such as a rejected token, never opens the app.SNIPAYA_MCP_NO_LAUNCH=1turns this off.The bridge only forwards. It runs no tools itself, stores nothing, and connects only to Snipaya. While Snipaya is not reachable, it answers the tool and prompt lists from a catalog shipped in the package. All actions run inside Snipaya, which shows connected assistants and recent activity in Settings > AI Assistants.
Tools
The tools come from the Snipaya version you have installed. Until Snipaya answers, the bridge lists the tools of Snipaya 1.3.3 from its catalog, and once Snipaya answers, the list of your installed version applies. Snipaya 1.3 provides:
Tool | Title |
| Get canvas |
| Render canvas |
| Add image |
| Add arrow or line |
| Add shape |
| Add text |
| Add step number |
| Add emoji |
| Add freehand stroke |
| Add magnifier |
| Add callout |
| Add spotlight |
| Add measurement |
| Pick color |
| Read text |
| Redact sensitive data |
| Update elements |
| Delete elements |
| Select elements |
| Arrange images |
| Set crop |
| Undo |
| Redo |
| Manage tabs |
| Export image |
| List export destinations |
| Export to destination |
| Save project |
Prompts
Prompt | Title | Optional argument |
| Explain a bug in a screenshot |
|
| Redact secrets before sharing |
|
| Make a screenshot look finished |
|
| Copy text out of an image |
|
| Turn screenshots into a walkthrough |
|
| Compare before and after |
|
Troubleshooting
The bridge writes the problem to stderr (never stdout, which carries the protocol). While Snipaya is not reachable, it still lists the tools, and returns the same text as the result of each tool call. When Snipaya starts answering, for example after you open it or accept the license agreement, the next tool call reaches it without a restart.
Message | Fix |
| Install Snipaya, open it once and turn on Allow AI assistants. Or set |
| The bridge runs on another system. It lists the tools, but they work only with Snipaya on a Mac. |
| Turn on Allow AI assistants in Settings > AI Assistants. |
| Open Snipaya, which rewrites its settings, then check Allow AI assistants. Or set |
| Snipaya is not installed in a place macOS can find. Install it from snipaya.app or open it yourself. |
| Snipaya 1.1 and newer start the MCP server only after you accept the license agreement. Accept it in the Snipaya window, then call the tool again. |
| Open Settings > AI Assistants and fix the error shown under the status. |
| Open Snipaya and turn on Allow AI assistants. If the message mentions the license agreement, accept it in Snipaya first. If the status shows an error, fix that first. Check that the URL matches Server URL. |
| The token changed, for example after Regenerate. Remove |
| Snipaya restarted or the token changed. Restart the snipaya server in your client. |
| The URL must point to |
| The bridge never started: a project |
Run npx -y snipaya-mcp --help to see the options.
Development
npm ci
npm run lint
npm run typecheck
npm test # builds, then runs the bridge over stdio against a local test server
npm run pack:mcpb # builds snipaya-mcp-<version>.mcpb with the official mcpb CLICatalog
catalog.json holds the tools, prompts and instructions the bridge serves while Snipaya is not reachable, and manifest.json lists the same tools and prompts. Regenerate both from a checkout of the Snipaya app (default ../snipaya) with Node 22.6 or newer, and commit the result:
npm run sync:catalog -- ../snipayaWithout a token or URL, the bridge reads the settings of the installed Snipaya in ~/Library/Application Support/Snipaya. Development builds of Snipaya keep their settings in their own Snipaya Dev profile folder, so to bridge to one, set SNIPAYA_MCP_TOKEN and SNIPAYA_MCP_URL to the values shown in its Settings > AI Assistants.
Releasing
Pushing a v* tag runs .github/workflows/release.yml. It runs the CI checks, checks that the tag matches the version in package.json, package-lock.json, server.json and manifest.json, stages the version on npm (with provenance), waits up to 60 minutes for a maintainer to approve it, then creates the GitHub release with the .mcpb attached, and publishes server.json to the MCP Registry.
One-time setup
On npmjs.com, open snipaya-mcp > Settings > Trusted publishing, choose GitHub Actions and enter:
Field | Value |
Organization or user |
|
Repository |
|
Workflow filename |
|
Environment name | leave empty |
Allowed actions |
|
npm expires a trusted publisher that has not published within 2 days, so add it shortly before the first release. The MCP Registry and the GitHub release need no setup: they use the workflow's own GitHub token.
Cut a release
npm version minor # or patch, major: bumps all four files, commits and tags v<version>
git push --follow-tagsWhen the workflow reaches Publish to npm, approve the staged version with a 2FA prompt: npm stage approve snipaya-mcp@<version>, or on the package page on npmjs.com. The rest of the release continues on its own.
Re-running a failed release is safe: it skips the npm and MCP Registry publish when that version is already there, and replaces the bundle on an existing GitHub release.
License
Available Tools
28 toolssnipaya_add_arrowAdd arrow or lineB
Draw a straight arrow (head at "to") or, with head false, a line. curved true draws a gentle curved arrow. Ends landing on an image are anchored to it.
Inputs: to (required), from (required), head, style, +3 more.
snipaya_add_calloutAdd calloutB
Add a speech-bubble callout with text in "rect" and a tail pointing at "tail" (defaults to below the box). The box grows to fit the text.
Inputs: rect (required), text (required), tail, style, +2 more.
snipaya_add_emojiAdd emojiB
Place an emoji sticker centered at "at".
Inputs: at (required), emoji (required), size, tabId, +1 more.
snipaya_add_freehandAdd freehand strokeB
Draw a freehand pen line, highlighter stroke or curved arrow through the given points. The path is smoothed like a hand-drawn stroke.
Inputs: kind (required), points (required), style, tabId, +1 more.
snipaya_add_imageAdd imageA
Add a screenshot or image to the canvas from an absolute file path or a data URL. Related images (before and after, steps of a flow) go on one canvas with placement "add". Use "new-tab" only to start a new, unrelated image, never one tab per related image.
Inputs: path, dataUrl, placement.
snipaya_add_magnifierAdd magnifierA
Add a round magnifying lens at "lens" that shows a zoomed view of the area around "source". radius defaults to about 8% of the short side of the image under the source.
Inputs: lens (required), source (required), zoom, style, +3 more.
snipaya_add_measureAdd measurementA
Draw a measuring line from "from" to "to" labeled with its length in image pixels. Returns the label.
Inputs: to (required), from (required), style, tabId, +1 more.
snipaya_add_shapeAdd shapeA
Add a rectangle frame, ellipse, highlighter box, blur (pixelate) or noise (opaque random colors from the area, the safest way to hide sensitive data) over a region. For blur and noise, pixelSize sets the block size; larger is coarser.
Inputs: kind (required), rect (required), style, tabId, +2 more.
snipaya_add_spotlightAdd spotlightA
Dim everything except "rect" (form rect or ellipse) to draw attention to it. dim (0 to 1) changes the user’s spotlight darkness setting for all spotlights.
Inputs: rect (required), dim, form, tabId, +1 more.
snipaya_add_stepAdd step numberA
Add an auto-numbered step badge (1, 2, 3...) centered at "at", for step-by-step guides. size is the badge diameter; make it larger on big screenshots.
Inputs: at (required), size, style, tabId, +1 more.
snipaya_add_textAdd textA
Add a text label with its top-left corner at "at". Supports \n for new lines.
Inputs: at (required), text (required), style, tabId, +1 more.
snipaya_arrangeArrange imagesB
Lay out all images of the tab in a row, column or grid with an optional gap.
Inputs: layout (required), gap, tabId.
snipaya_delete_itemsDelete elementsADestructive
Delete elements (images or annotations) by id. One undo step.
Inputs: ids (required), tabId.
snipaya_exportExport imageADestructive
Export the canvas (or given ids) as PNG or JPG. With path, writes the file (absolute path, matching extension, refuses to overwrite unless overwrite is true); without path, returns the image.
Inputs: format (required), ids, path, tabId, +2 more.
snipaya_export_toExport to destinationADestructive
Render the canvas (or given ids) as PNG and send it to a destination from snipaya_list_destinations. Uploads to network destinations always ask the user to confirm first. Returns the link or file path.
Inputs: destination (required), ids, tabId, beautify.
snipaya_get_canvasGet canvasARead-only
Describe the open tabs and the active canvas: images (id, position, size, original pixel size), annotations (id, kind, geometry, style, text, anchors), selection and crop. Call this first.
Inputs: tabId.
snipaya_list_destinationsList export destinationsARead-only
List the export destinations the user set up (share menu, open with an app, quick save folder, uploaders, Jira, Confluence) with their ids for snipaya_export_to.
Inputs: no named parameters.
snipaya_pick_colorPick colorARead-only
Read the color of the image pixel at "at" and return it as a #rrggbb hex. Changes nothing.
Inputs: at (required), tabId, relativeTo.
snipaya_read_textRead textARead-only
Recognize text (on-device OCR) in the images and return the lines in reading order with their rects. Limit it with rect and/or imageId. Rects use the relativeTo space when set.
Inputs: rect, tabId, imageId, relativeTo.
snipaya_redactRedact sensitive dataA
Find emails, phone numbers, IBANs, card numbers, tokens, IP addresses, passwords and faces (on-device) and cover them in one undo step, in the user’s Redaction style (noise by default). style overrides it: noise fills the area with opaque random colors from it, blur pixelates with blocks sized to the text, solid covers it with a plain box. pixelSize sets the noise block size or a coarser minimum for blur. Defaults to every image and category. Returns the ids and counts.
Inputs: style, tabId, imageIds, pixelSize, +1 more.
snipaya_redoRedoA
Redo the last undone change on the tab.
Inputs: tabId.
snipaya_renderRender canvasARead-only
Return a PNG of the canvas (or only the given element ids) so you can see what the user sees. Downscaled to maxSize on the longest side.
Inputs: ids, tabId, maxSize, beautify.
snipaya_save_projectSave projectADestructive
Save the tab as a .snipaya project. Without path, saves to the tab’s existing file; a tab that was never saved needs an absolute path.
Inputs: path, tabId, overwrite.
snipaya_selectSelect elementsA
Select elements by id so the user sees which ones you mean. An empty list clears the selection.
Inputs: ids (required), tabId.
snipaya_set_cropSet cropA
Crop exports to a rectangle, or pass null to remove the crop.
Inputs: rect (required), tabId, relativeTo.
snipaya_tabsManage tabsADestructive
Open a new empty tab, switch to a tab, or close a tab (closing never discards unsaved work without asking the user).
Inputs: action (required), tabId.
snipaya_undoUndoA
Undo the last change on the tab (yours or the user’s).
Inputs: tabId.
snipaya_update_itemsUpdate elementsA
Change existing elements by id: move (x, y), resize (width, height), rotate (degrees), change text, or restyle. For steps, width or style.fontSize resizes the badge. All updates are one undo step.
Inputs: updates (required), tabId, relativeTo.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 28 tools
Most tools have clearly distinct purposes: annotation creation, selection, arrangement, export, and canvas inspection are separated well. A few boundaries are adjacent rather than ambiguous, such as render vs export/export_to, add_shape blur/noise vs redact, and delete_items vs undo. The descriptions generally resolve these, so an agent can choose correctly.
All tools use the snipaya_ prefix with snake_case, which is highly consistent. However, naming is not a strict verb_noun pattern throughout: many annotation tools use add_, but there are also verb-only names like redact, render, select, arrange, undo, redo, export, and the noun-like tabs. Still, the convention is predictable and readable.
The 28-tool surface is heavy, but the domain is feature-rich image annotation and covers many distinct annotation primitives. Each tool has a plausible place, yet the set could likely be consolidated with a more generic annotation tool rather than exposing every annotation type separately. The count is borderline for an MCP server but not chaotic.
The server covers the annotation lifecycle well: inspect canvas, add diverse annotation types, read/select/update/delete elements, undo/redo, manage tabs, and export/save. Minor gaps remain, notably no obvious open/load project tool despite save_project, and no destination-management tool beyond listing existing destinations. These are workable gaps rather than dead ends for core annotation workflows.
Maintenance
Related MCP Connectors
Talk to your public-facing AI from any MCP client — Claude, ChatGPT, Cursor, Cline, Windsurf.
Focused MCP server for OpenAI image/audio generation (v2.0.0). Wraps endpoints via HAPI CLI.
MCP server for Clipkit — gives AI agents a video toolbox via the Clipkit schema.
Shared memory MCP server for Claude Code, Codex, Gemini CLI and Cursor. Cited, editable.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceA lightweight stdio MCP server that adds image understanding to AI coding assistants via a single tool that sends images to any OpenAI-compatible multimodal endpoint.MIT
- AlicenseAqualityCmaintenanceA local (stdio) MCP server for PixelVault — agent-first image hosting, enabling upload of local files by path without base64.613 npmMIT

mind-elixir-mcpofficial
AlicenseNot gradedqualityCmaintenanceBridges stdio-only MCP clients to the Mind Elixir Desktop app, enabling AI assistants to create, inspect, edit, and annotate mind maps in real time.65 npm1MIT- FlicenseNot gradedqualityCmaintenanceEnables local programs to push text and images to AI assistants via HTTP and retrieve AI-generated responses, acting as a local data bridge between your code and MCP-capable AI tools.-