# com.sketchxflow/mcp (npm · @sketchxflow/mcp) Design websites, apps and logos with SketchXFlow — and brief its development team. - Trust score: 62/100 (medium) - Registry status: active - Liveness: live - Owner verified: no - Last scored: 2026-09-26 ## Components - npm · `@sketchxflow/mcp`: 62/100 (this document), [markdown](https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp.md), [page](https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp) ## Channel facts - Registry: `npm` - Package: `@sketchxflow/mcp` - Version: `0.22.0` - Transport: `stdio` ## Trust breakdown 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. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog) Scored 2026-09-26. - **Supply Chain Security**: 98/100 - No malware found by supply-chain analysis. - No known CVEs affecting this package version or its production dependencies. - No install/post-install scripts declared. - 33 of 99 dependencies flagged as unhealthy. - **Provenance & Transparency**: 19/100 - Repository check failed: the declared repository URL redirects; it must resolve directly. - Provenance check failed: no build-provenance attestation is published. - Clear OSI-approved license (MIT). - Actively maintained (last published 6 days ago). - Security-disclosure policy not yet verified: we couldn't inspect the source repository. - **Schema Quality & AI Usability**: 59/100 - AI-judged instruction clarity (excellent). - Context-footprint check failed: tool/resource definitions use about 10477 tokens (~349/item across 30 items; 30 tools + 0 resources), over budget; trim descriptions and params. - Usage-examples check failed: none of the tools include examples. - **Stability & Change Management**: 0/100 - Stability not yet verified: not enough scan history yet (needs a 30-day window). - **Tool Coverage**: 96/100 - 100% of tools have a non-trivial description (not blank, and not just the tool's name). - 89% of tool parameters carry a description. - **Tool Safety**: 100/100 - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured. - We read all 30 captured tool definition(s), and no name or description among them implies an irreversible operation. - An AI judge read all 30 captured unit(s) of tool text and found none that tries to manipulate the model reading it. - **Capabilities**: 100/100 - Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28. **Unverified: 1 category.** A category scored 0 because we could not verify it: a data source with nothing on this package, evidence we could not reach, or a check we could not run. We only credit what we can confirm. ## Install ### How do I install the com.sketchxflow/mcp server? com.sketchxflow/mcp runs locally as an npm package, launched with npx -y @sketchxflow/mcp. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation. ### Claude ```bash claude mcp add com-sketchxflow-mcp -- npx -y @sketchxflow/mcp ``` ### Cursor ```json { "mcpServers": { "com-sketchxflow-mcp": { "command": "npx", "args": [ "-y", "@sketchxflow/mcp" ] } } } ``` ### VS Code ```json { "servers": { "com-sketchxflow-mcp": { "command": "npx", "args": [ "-y", "@sketchxflow/mcp" ] } } } ``` ### Codex ```bash codex mcp add com-sketchxflow-mcp -- npx -y @sketchxflow/mcp ``` ### opencode ```json { "$schema": "https://opencode.ai/config.json", "mcp": { "com-sketchxflow-mcp": { "type": "local", "command": [ "npx", "-y", "@sketchxflow/mcp" ], "enabled": true } } } ``` ### OpenClaw ```bash openclaw mcp add com-sketchxflow-mcp --command npx --arg -y --arg @sketchxflow/mcp ``` ### Hermes ```yaml mcp_servers: com-sketchxflow-mcp: command: "npx" args: ["-y", "@sketchxflow/mcp"] ``` ### Netclaw ```json { "McpServers": { "com-sketchxflow-mcp": { "Transport": "stdio", "Command": "npx", "Arguments": [ "-y", "@sketchxflow/mcp" ] } } } ``` ### Vellum ```bash assistant mcp add com-sketchxflow-mcp -t stdio -c npx -a -y @sketchxflow/mcp ``` ### Other ```json { "mcpServers": { "com-sketchxflow-mcp": { "command": "npx", "args": [ "-y", "@sketchxflow/mcp" ] } } } ``` ## Changelog Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims. ### 2026-09-25 (score 62, 0) - [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server ### 2026-09-20 (score 62) First indexed and scored. ## MCP tools (30) ### `sketchxflow_version` (~83 tokens) Check the SketchXFlow MCP version and API key Report which version of the SketchXFlow MCP server is running, whether a newer one is published, and whether the SketchXFlow API accepts your SKETCHXFLOW_API_KEY (shown only by its sk_live_ab12cd… prefix). Free, instant, no coins. Call it when another tool fails with an API-key error, or to check for updates. ### `sketchxflow_list_models` (~150 tokens) List the LLM models you can pick List the models you can pick — BOTH catalogs, kept separate: the LLM TEXT models for the `model` parameter (the model that writes the page) and the IMAGE models for `image_model` (the model that renders the pictures in it). Each entry carries a description, speed, relative cost, coin rates, and whether YOUR plan can actually use it. Also reports your plan, the defaults used when the parameters are omitted, and how to unlock selection if it's locked. The two id lists do not overlap and are not interchangeable. Free, instant, no coins. Call this before passing `model` or `image_model`, or to answer 'which models can I use?'. ### `sketchxflow_clarify` (~390 tokens) Ask the user design questions Use BEFORE designing when the brief is thin. Send the user's latest message (and the running `vision` + `history` from prior turns); get back a reply to relay, the updated `vision` board, `ready` (true when there's enough to design), and `missing` topics. When ready, the response includes `design_prompt` — pass that to sketchxflow_design_page. Free. Loop this with the user one question at a time. Input parameters: - `history` (array): Prior turns (echo back what you got last call). - `message` (string, required): The user's latest answer / the initial idea. - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `platform` (string) - `vision`: The `vision` object returned last call — pass it back each turn. ### `sketchxflow_design_page` (~763 tokens) Design a page (new project or add to one) Generate ONE polished page. Omit project_id to start a NEW project (creates the project + its first page). Pass an existing project_id to ADD a page to that project (brand-consistent). IMPORTANT: if the user wants a MOBILE app or phone screen (iOS/Android/app UI), you MUST set platform='mobile' — it defaults to a web/desktop design otherwise. Returns a preview URL; call sketchxflow_get_code to pull the HTML to implement. Takes 1–9 min; streams progress. Input parameters: - `ai_images` (boolean) - `build_3d` (boolean): Commission a bespoke 3D model in the user's OWN Blender instead of casting one from the stock library. Requires enable_3d AND a machine paired via Account -> Machines with the SketchXFlow bridge runn… - `enable_3d` (boolean) - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `name` (string): Page name/title (used when adding to a project, e.g. 'Pricing'). Defaults from the prompt. - `platform` (string): Target surface for a NEW project. Set 'mobile' for a mobile app / phone screen / iOS / Android / app UI; 'web' for websites and desktop. Defaults to web — so if the user asked for anything mobile, yo… - `project_id` (string): Existing project to add this page to. Omit to create a new project. - `prompt` (string, required): What this page is — the brief. ### `sketchxflow_plan` (~526 tokens) Plan a multi-page site (before generating) Create the project and produce its SITE PLAN — the page list (slug, name, purpose) and each page's section breakdown — for review BEFORE any HTML is generated. Fast and synchronous; the plan is saved server-side against the returned project_id. Show it to the user, adjust with sketchxflow_refine_plan, then build with sketchxflow_generate_from_plan. WHEN TO USE: only for MULTI-PAGE sites (roughly 3+ pages). Planning first produces markedly better multi-page results — the model gets the whole site map before it writes any page, so navigation, tone and structure stay consistent across them. For a SINGLE page it is the wrong call: it is two extra round trips and an extra coin charge for no gain, and sketchxflow_design_page alone is the right tool. NOT a substitute for sketchxflow_clarify, and not an alternative to it: clarify INTERVIEWS THE USER to build a brief and produces no plan; these tools take a finished brief and produce a site map. The full sequence when the brief is thin is clarify (loop with the user) → plan → refine_plan → generate_from_plan. Input parameters: - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `platform` (string): Target surface. 'mobile' for a mobile app / phone screens; defaults to web. - `prompt` (string, required): The brief for the whole site — what it is, who it's for, and any pages you already know it needs. ### `sketchxflow_refine_plan` (~533 tokens) Refine a saved site plan Apply one natural-language change to a project's saved site plan and return the updated plan — 'add a careers page', 'drop the blog', 'the pricing page needs a comparison table', 'reorder so About comes before Pricing'. Fast and synchronous; nothing is built yet, so this is the cheap place to iterate. Loop it with the user until the page list is right, then call sketchxflow_generate_from_plan. Requires a plan created by sketchxflow_plan. WHEN TO USE: only for MULTI-PAGE sites (roughly 3+ pages). Planning first produces markedly better multi-page results — the model gets the whole site map before it writes any page, so navigation, tone and structure stay consistent across them. For a SINGLE page it is the wrong call: it is two extra round trips and an extra coin charge for no gain, and sketchxflow_design_page alone is the right tool. NOT a substitute for sketchxflow_clarify, and not an alternative to it: clarify INTERVIEWS THE USER to build a brief and produces no plan; these tools take a finished brief and produce a site map. The full sequence when the brief is thin is clarify (loop with the user) → plan → refine_plan → generate_from_plan. Input parameters: - `instruction` (string, required): The change to make to the plan (pages, their purpose, or their sections). - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `project_id` (string, required): The project returned by sketchxflow_plan. ### `sketchxflow_generate_from_plan` (~693 tokens) Build the site from its saved plan Generate every page of a project's saved site plan, in one consistent pass. This is the step that costs coins and takes time (a few minutes, scaling with the page count); streams progress. Call it once the user is happy with the plan from sketchxflow_plan / sketchxflow_refine_plan — building is where changes get expensive, so settle the plan first. Then pull each page's HTML with sketchxflow_get_code. WHEN TO USE: only for MULTI-PAGE sites (roughly 3+ pages). Planning first produces markedly better multi-page results — the model gets the whole site map before it writes any page, so navigation, tone and structure stay consistent across them. For a SINGLE page it is the wrong call: it is two extra round trips and an extra coin charge for no gain, and sketchxflow_design_page alone is the right tool. NOT a substitute for sketchxflow_clarify, and not an alternative to it: clarify INTERVIEWS THE USER to build a brief and produces no plan; these tools take a finished brief and produce a site map. The full sequence when the brief is thin is clarify (loop with the user) → plan → refine_plan → generate_from_plan. Input parameters: - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `project_id` (string, required): The project whose saved plan should be built. ### `sketchxflow_design_creative` (~805 tokens) Design a creative (flyer / poster / social) Generate a fixed-canvas graphic — a flyer, poster, social post/story, or an AD creative for Meta / Instagram / Google (feed, story, reel, leaderboard, rectangle, skyscraper, mobile banner, billboard…) — from a brief. Pick the placement via `format`. Returns a preview URL and a rendered PNG. This is for STATIC creatives, distinct from sketchxflow_design_page (which builds websites/app screens). Takes ~1–3 min; streams progress. Refine it with sketchxflow_edit and pull markup with sketchxflow_get_code (slug 'home'). Input parameters: - `ai_images` (boolean): Generate AI imagery (default true). false = type/color/shape only. - `format` (string): Canvas format. Default instagram_post (1080×1080). General: instagram_story 1080×1920, poster_a4 print, flyer, x_post/linkedin_post/youtube_thumbnail landscape. AD FORMATS — Meta: meta_feed_landscape… - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `prompt` (string, required): The creative brief — what it's for, the message/headline, the vibe, any details (date, price, handle, URL). ### `sketchxflow_design_email` (~623 tokens) Design an email template (HTML email) Generate a responsive, client-safe HTML email — newsletter, welcome, promotional, announcement, transactional, or event invite — from a brief. Uses email-safe rules (table layout, inline CSS, 600px column) so it renders in Gmail / Apple Mail / Outlook. Returns a preview URL and a rendered PNG; pull the HTML with sketchxflow_get_code (slug 'home'). Distinct from sketchxflow_design_page (websites) and sketchxflow_design_creative (static graphics). Takes ~1–3 min; streams progress. Input parameters: - `ai_images` (boolean): Generate AI imagery (default true). false = type/color/layout only. - `email_type` (string): Email type. Default newsletter. welcome = onboarding; promotional = offer/sale; transactional = receipt/confirmation; event = invite. - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `prompt` (string, required): The email brief — purpose, subject/headline, key message, CTA, and any details (offer, dates, links). ### `sketchxflow_launch_kit` (~594 tokens) Generate a launch kit (landing + email + socials + poster) Generate a complete, on-brand LAUNCH KIT from one brief — a landing page + launch email + social posts + poster, all consistent. Auto-creates a brand (palette/fonts) if none is attached. Returns the set of preview URLs + project IDs. Takes ~3–8 min; streams progress. Pull any asset's code with sketchxflow_get_code (slug 'home'). Input parameters: - `ai_images` (boolean): Generate AI imagery (default true). false = type/color/layout only. - `brand_kit_id` (string): Attach a saved brand kit (see sketchxflow_brand_kits). Omit to auto-create + save a brand from the brief. - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `prompt` (string, required): The launch brief — what's launching, who it's for, the message, and any details (date, offer, link). ### `sketchxflow_create_logo` (~619 tokens) Generate a brand logo Design a brand LOGO from a name + brief — a symbol mark plus wordmark, lockups, and favicons. Optionally ground it on an existing project so it inherits that brand's palette/fonts. Async; streams progress (~1–3 min). Returns a displayable logo image URL you can drop into a design with sketchxflow_add_image(image_url=…). Input parameters: - `description` (string): What the brand does — one or two sentences. Grounds the symbol concept. - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `name` (string, required): The brand name to put on the logo (required). - `source_project_id` (string): Ground the logo on this existing project's palette/fonts/brand so it matches the site you already designed. - `vibes` (array): Visual vibe chips, e.g. ['minimal','bold','technical','warm','playful','refined','premium','earthy']. - `website` (string): Brand website to research for grounding (optional). ### `sketchxflow_create_animation` (~552 tokens) Generate an animation / product-demo video Generate a motion-graphics ANIMATION from a brief — a branded intro / logo reveal / kinetic-typography sting, or a product demo. It becomes a self-contained, auto-playing motion page (timeline JSX). Optionally GROUND it on an existing project (`source_project_id` — its pages render to REAL screenshots that animate into the video), a logo (`source_logo_id`), and/or a live URL. Async (~1–4 min); the result preview_url plays the animation. MP4 export is coming; for now share the preview URL. Input parameters: - `brand_kit_id` (string): Attach a saved brand kit (see sketchxflow_brand_kits). - `demo` (boolean): Set true for a PRODUCT DEMO VIDEO: the attached screenshots (from source_project_id pages / source_url / images) are VISION-UNDERSTOOD and turned into a multi-scene demo (hook → feature scenes with r… - `duration` (number): Length in seconds (3–15, default 6). - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `prompt` (string, required): The animation brief — what it's for, the message/vibe, what to highlight. - `source_logo_id` (string): Feature one of your created logos in the animation. - `source_project_id` (string): Ground on this existing project — its pages render to real screenshots that animate in (great for product demos). - `source_url` (string): A live product URL to screenshot and feature. ### `sketchxflow_brand_kits` (~45 tokens) List your saved brand kits List the brand kits (saved brand identities — name, palette, fonts, voice) in your account, to attach to a launch kit via brand_kit_id. ### `sketchxflow_edit` (~336 tokens) Edit a page section/component Apply a natural-language change to one page — e.g. 'make the hero darker', 'tighten the pricing section', 'swap the testimonial for a logo wall'. Brand-locked: edits stay on-palette/on-type. Returns the updated page; call sketchxflow_get_code for the new HTML. Input parameters: - `instruction` (string, required): The change to make. - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `project_id` (string, required) - `slug` (string, required): Which page (e.g. 'home', 'pricing'). ### `sketchxflow_get_code` (~83 tokens) Get a page's HTML to implement Return the full HTML (Tailwind inline) of a generated page so you can re-implement or adapt it into the user's codebase (React/Next/Vue/plain HTML). Call after sketchxflow_design_page / sketchxflow_edit. Input parameters: - `project_id` (string, required) - `slug` (string, required): Page slug, e.g. 'home'. ### `sketchxflow_get_image` (~97 tokens) See a page's rendered design Return a rendered screenshot (PNG) of a page so you can SEE how the design actually looks. Use it to visually verify or critique a design after sketchxflow_design_page / sketchxflow_design_creative / sketchxflow_edit, before implementing or iterating. slug 'home' is the main page / creative. Input parameters: - `project_id` (string, required) - `slug` (string): Page slug. Default 'home'. ### `sketchxflow_add_image` (~482 tokens) Add an image to a design (poster / creative / page) Insert an image into an existing design — a poster, creative, or any page. Either AI-GENERATE one from `prompt` (costs coins, respects the project's AI-image toggle) OR drop in an existing image by `image_url` (a public http(s) URL). Optional `placement` positions it ('top right', 'center', 'bottom left', …) — it goes in as a movable overlay you can reposition/resize afterward with sketchxflow_edit. Then view the result with sketchxflow_get_image. Input parameters: - `alt` (string): Alt text for the image. - `height` (number): Generated image height in px (default 600). Only used with prompt. - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `image_url` (string): Public http(s) URL of an existing image to insert. Provide this OR prompt. - `placement` (string): Where to place it: top left / top right / bottom left / bottom right / top / bottom / left / right / center. Default center. - `project_id` (string, required): The project to add the image to. - `prompt` (string): Describe an image to AI-generate (costs coins). Provide this OR image_url. - `slug` (string): Page slug. Default 'home' (the main page / creative). - `width` (number): Generated image width in px (default 800). Only used with prompt. ### `sketchxflow_experimental_start` (~615 tokens) Start a layered design (Experimental) Open an EXPERIMENTAL DESIGN session: one surface, built the way a designer builds one — a SECTION at a time, each section STACKED up a layer at a time (a ground plane, shapes over it, the copy, imagery, effects), then whole-page passes at the end to make the sections agree. Returns the SECTION PLAN plus the art director's opening reply and the first layer proposal, for you to relay to the user. Free and instant; nothing is rendered and no coins are spent until sketchxflow_experimental_layer. WHEN TO USE: for ONE surface (a page or a screen) the user wants to shape WITH you section by section rather than receive finished — 'let's design this together', 'I want to iterate on the look', 'build it up and show me'. For a page they just want generated, sketchxflow_design_page is one call and is the right tool; for a multi-page site, use sketchxflow_plan. This is a CONVERSATION: relay the director's reply to the user, get their answer, and pass it back — do not run the whole stack unless they have said to. Input parameters: - `auto` (boolean): Put the director on auto-pilot from the start — it stops asking and fills the layers in itself. Set this ONLY if the user has said they want it built without being asked each time. - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `platform` (string): Target surface. 'mobile' for a phone screen / app UI; defaults to web. - `project_id` (string): Layer on top of an EXISTING project instead of creating one. Omit for a new design. - `prompt` (string, required): The brief — what this design is, who it's for, and any direction the user has already given. - `slug` (string): Which page of the project to build. Defaults to 'home'. ### `sketchxflow_experimental_discuss` (~526 tokens) Talk over the next layer (Experimental) One turn with the art director WITHOUT applying anything: pass what the user said and get back a revised proposal for the next layer. Use it when they want to change direction, ask a question, or think out loud before committing — 'warmer than that', 'what would you do instead?', 'can we skip the imagery for now?'. Free, instant, no coins. Loop this with the user, then apply with sketchxflow_experimental_layer. WHEN TO USE: for ONE surface (a page or a screen) the user wants to shape WITH you section by section rather than receive finished — 'let's design this together', 'I want to iterate on the look', 'build it up and show me'. For a page they just want generated, sketchxflow_design_page is one call and is the right tool; for a multi-page site, use sketchxflow_plan. This is a CONVERSATION: relay the director's reply to the user, get their answer, and pass it back — do not run the whole stack unless they have said to. Input parameters: - `auto` (boolean): Turn auto-pilot on or off. Only set true if the user explicitly asked for the rest to be filled in without being asked each time. - `message` (string, required): What the user just said, in their words. - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `project_id` (string, required): The layered session (from sketchxflow_experimental_start). ### `sketchxflow_experimental_layer` (~1026 tokens) Apply the next layer (Experimental) Stack ONE layer onto the design and return what it put down plus the director's next proposal. This is the step that costs coins. A SECTION layer is small and quick (a few seconds, ~3 coins) because it rewrites only that section; a whole-page finishing pass rewrites the document and costs more. The very first call always STAGES the page (the section shells) whatever else you pass. Pass `steer` to fold the user's words into the pending layer, or `kind` + `instruction` to say exactly what this layer is. `count` runs several layers back to back WITHOUT stopping to ask: only pass it when the user has explicitly said to keep going (e.g. 'just build the rest'), because each layer in the run is charged. After it returns, relay the next proposal to the user. WHEN TO USE: for ONE surface (a page or a screen) the user wants to shape WITH you section by section rather than receive finished — 'let's design this together', 'I want to iterate on the look', 'build it up and show me'. For a page they just want generated, sketchxflow_design_page is one call and is the right tool; for a multi-page site, use sketchxflow_plan. This is a CONVERSATION: relay the director's reply to the user, get their answer, and pass it back — do not run the whole stack unless they have said to. Input parameters: - `count` (integer): Apply this many layers in one run without stopping to ask. Requires the user's explicit go-ahead — every layer in the run is charged. Defaults to 1. - `image_model` (string): Optional. Pin which model generates this design's AI images — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific image model: omitting uses the account's default (g… - `instruction` (string): Replace the director's proposal with your own brief for this layer. Be specific — hexes, geometry, placement in the stack. Changes exactly what it names and nothing else. - `model` (string): Optional. Pin which LLM generates this — a PAID-PLAN feature (Starter and above). OMIT IT unless the user asked for a specific model: omitting uses the account's default (gemini-3-flash-preview) and… - `project_id` (string, required): The layered session (from sketchxflow_experimental_start). - `role` (string): What KIND of layer this is. Only with `instruction`. SECTION roles: backdrop=the ground plane, shape=geometry over it, content=the copy, media=imagery, effect=grain/glass/glow. PAGE passes: palette,… - `scope` (string): 'section' (default) stacks a layer inside ONE section. 'page' runs a whole-page finishing pass — only worth doing once every section is built. - `section_id` (string): Which section to stack into, e.g. 'hero'. Required when you pass `instruction` with scope='section'. Read the ids from sketchxflow_experimental_stack. - `steer` (string): The user's steer on the pending layer, folded into its brief — 'warmer, less orange', 'push the panel further left'. - `title` (string): A short name for the layer, when you're specifying it yourself. ### `sketchxflow_experimental_stack` (~322 tokens) Read or peel back the layer stack (Experimental) Read a layered design's build: which sections exist, what is stacked inside each one, which whole-page passes have run, and what the director proposes next. Use it to get the section ids before targeting a layer. Pass `revert_to` to PEEL LAYERS BACK OFF — `revert_to=3` keeps the first three and discards everything above them, restoring the canvas to exactly how it looked then (`revert_to=0` clears it entirely). Reverting is free and instant, because each layer was snapshotted when it was applied — but it DISCARDS those layers, so confirm with the user before passing it. Omit it to just read. WHEN TO USE: for ONE surface (a page or a screen) the user wants to shape WITH you section by section rather than receive finished — 'let's design this together', 'I want to iterate on the look', 'build it up and show me'. For a page they just want generated, sketchxflow_design_page is one call and is the right tool; for a multi-page site, use sketchxflow_plan. This is a CONVERSATION: relay the director's reply to the user, get their answer, and pass it back — do not run the whole stack unless they have said to. Input parameters: - `project_id` (string, required): The layered session. - `revert_to` (integer): Keep this many layers and discard the rest. OMIT to read without changing anything. Confirm with the user first — the discarded layers are gone. ### `sketchxflow_check_job` (~136 tokens) Collect a design that was still building Pick up a generation that was still running when the tool call returned. Pass the job_id from a '⏳ Still …' reply; this waits for it and returns the finished result, or reports how far along it is. Safe to call repeatedly — it only reads, so it never charges again and never starts a second run. This is the RIGHT response to a '⏳ Still …' reply; re-sending the original request is not, because that starts a second job and charges for it. Input parameters: - `job_id` (string, required): The job_id from a '⏳ Still building' reply. ### `sketchxflow_get_project` (~31 tokens) Get a project's status Project status: preview URL + page list (slug/name). Input parameters: - `project_id` (string, required) ### `sketchxflow_list_projects` (~25 tokens) List your projects List the projects in your SketchXFlow account (most recent first). ### `sketchxflow_services_projects` (~38 tokens) List your development-services projects Your projects with SketchXFlow's development services — the ones you've asked the team to build — with their state and links. Free. ### `sketchxflow_services_start` (~119 tokens) Start a development-services project Ask SketchXFlow to BUILD something for real (a website, web app or mobile app) at a fixed price from the public price list. Opens a project and returns the AI consultant's first question. Free; nothing is committed until a proposal is accepted on the website. Input parameters: - `brief` (string): What you need built, in your own words — who it's for and what it should do. Paste an existing brief if you have one. - `title` (string): A short working title, e.g. 'Clinic booking site'. ### `sketchxflow_services_say` (~97 tokens) Answer the SketchXFlow consultant One turn of the requirements interview: send what the client said, get the consultant's reply, the question they should answer next (with options if any) and the live estimate from the price list. Free. Keep going until can_finalise is true, then call sketchxflow_services_finalise. Input parameters: - `message` (string, required): What the client said, or the option they picked. - `project_id` (string, required) ### `sketchxflow_services_requirements` (~45 tokens) Read a project's requirements document The requirements document the consultant is writing (sections, numbered requirements, modules), plus whether it can be finalised. Free. Input parameters: - `project_id` (string, required) ### `sketchxflow_services_finalise` (~63 tokens) Finalise the requirements and get the proposal Freeze the requirements and price them from the price list. Only with the client's explicit go-ahead. Returns whether the proposal issued and the link to read, accept and pay it on the website (sign-in needed). Input parameters: - `project_id` (string, required) ### `sketchxflow_services_invoices` (~60 tokens) Invoices and payment links for a project Every invoice on a project with its status and, for an open one, the page to pay it (UPI, cards, EMI, net banking or bank transfer; sign-in needed). Free. Input parameters: - `project_id` (string, required) ## Diagnostics Captured diagnostic sections: Provenance, Dependencies. The full working is on the page: https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp#diagnostics ## Score history - 2026-09-26: 62 - 2026-09-25: 62 - 2026-09-24: 62 - 2026-09-23: 62 - 2026-09-22: 62 - 2026-09-21: 62 - 2026-09-20: 62 ## Common questions ### What is the com.sketchxflow/mcp server? com.sketchxflow/mcp is listed in the public MCP registry as com.sketchxflow/mcp. Design websites, apps and logos with SketchXFlow, and brief its development team. This page covers its npm package (@sketchxflow/mcp). ### Is the com.sketchxflow/mcp server safe to use? com.sketchxflow/mcp scores 62 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 26 September 2026. It declares no install or post-install scripts. 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 com.sketchxflow/mcp server expose? com.sketchxflow/mcp exposes 30 tools: sketchxflow_version, sketchxflow_list_models, sketchxflow_clarify, sketchxflow_design_page, sketchxflow_plan, and 25 more. Their descriptions and schemas cost roughly 10,477 tokens of context every time the server is loaded. ### Is the com.sketchxflow/mcp server still maintained? com.sketchxflow/mcp is still listed as active in the MCP registry. We last reached this channel on 26 September 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 com.sketchxflow/mcp server under? com.sketchxflow/mcp declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls. ## Links - npm package: https://www.npmjs.com/package/@sketchxflow/mcp - Socket report: https://socket.dev/npm/package/@sketchxflow/mcp - Website: https://sketchxflow.com/mcp - Changelog RSS feed: https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp.xml - Changelog JSON feed: https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp.json - HTML version of this page: https://verifymcp.io/servers/com-sketchxflow-mcp/sketchxflow-mcp