"A method to allow vision models to process and understand image files" matching MCP tools:

• get_fabrication_quote

  Agentic Fabrication Network (AFN)

  Get ranked, purchasable offers (price, ETA, preview image) for fabricating a physical item from a design file. process=fdm_print for 3D printing a model (STL/OBJ/PLY/3MF/AMF/STEP/IGES), process=cnc or process=sheetmetal for machined/bent metal parts (STEP, IGES, DXF), process=decal for stickers/decals from artwork (any common image or design file — PNG/JPG/HEIC/TIFF/GIF/BMP/WEBP/AVIF/SVG/PDF/AI/EPS/PSD/CDR, auto-converted). A .ufp file (UFP part container: the design plus saved spec/constraints in one) is accepted anywhere a design file is — its saved intent applies automatically and anything the user states now wins. If the user just drops a file and asks for a price, omit process — UFP routes it. Provide the design either as design_file (an image/file the user attached or you generated — preferred) or file_url (a public URL). REORDERS: if the user has a UFP part number (from a receipt email or a previous session, looks like UFP-… or part_…), pass it as part_number INSTEAD of any file — the stored design and spec are reused and re-shopped across all current vendors. Locked parts additionally require share_key (from the owner's share link). Returns offers across vendors like Google Flights returns flights.

  Connector

• upload_product_image

  Partle

  Attach an image to an existing product by giving Partle a public URL to download the image from. Authenticated. OAuth (scope `products:write`) preferred; `api_key` fallback. **When to use this tool**: the image is already hosted at a public URL (a scraped product page, an Imgur link, a CDN URL the user provided). Partle's server fetches it and stores it. **When NOT to use this tool**: you have local image bytes (a file the user attached, or bytes you generated/downloaded in your sandbox). Sending those bytes through a tool argument blows past conversation context limits — phone-photo-sized payloads can be 6+ MB of base64. Instead, in your code-execution sandbox, POST the file directly to the HTTP endpoint with multipart encoding: requests.post( "https://partle.rubenayla.xyz/v1/external/products/{product_id}/images", files={"file": open("/path/to/photo.jpg", "rb")}, headers={"X-API-Key": "pk_..."}, ) Or, to create the listing and attach an image in one HTTP request: requests.post( "https://partle.rubenayla.xyz/v1/external/products", data={"metadata": json.dumps({"name": ..., "price": ...})}, files={"image": open("/path/to/photo.jpg", "rb")}, headers={"X-API-Key": "pk_..."}, ) Args: product_id: ID of the product to attach the image to. image_url: Publicly fetchable URL of the image. Server fetches it and stores it. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: The created `ProductImage` record with its `id` (use for deletion) and storage path, or ``{"error": ...}`` on validation/auth failure.

  Connector

• upload_product_image

  Partle Marketplac

  Attach an image to an existing product by giving Partle a public URL to download the image from. Authenticated. OAuth (scope `products:write`) preferred; `api_key` fallback. **When to use this tool**: the image is already hosted at a public URL (a scraped product page, an Imgur link, a CDN URL the user provided). Partle's server fetches it and stores it. **When NOT to use this tool**: you have local image bytes (a file the user attached, or bytes you generated/downloaded in your sandbox). Sending those bytes through a tool argument blows past conversation context limits — phone-photo-sized payloads can be 6+ MB of base64. Instead, in your code-execution sandbox, POST the file directly to the HTTP endpoint with multipart encoding: requests.post( "https://partle.rubenayla.xyz/v1/external/products/{product_id}/images", files={"file": open("/path/to/photo.jpg", "rb")}, headers={"X-API-Key": "pk_..."}, ) Or, to create the listing and attach an image in one HTTP request: requests.post( "https://partle.rubenayla.xyz/v1/external/products", data={"metadata": json.dumps({"name": ..., "price": ...})}, files={"image": open("/path/to/photo.jpg", "rb")}, headers={"X-API-Key": "pk_..."}, ) Args: product_id: ID of the product to attach the image to. image_url: Publicly fetchable URL of the image. Server fetches it and stores it. api_key: Optional API key (`pk_*`, generate at /account). Used when there is no OAuth token, and also when the OAuth token lacks the required scope — an explicitly passed key overrides an ambient token that is scoped too narrowly. An invalid or revoked token still fails regardless. Omit when using OAuth. Returns: The created `ProductImage` record with its `id` (use for deletion) and storage path, or ``{"error": ...}`` on validation/auth failure.

  Connector

• squish_video

  Squish

  Turn a video at a public URL into timestamped contact-sheet JPEG(s) that a vision model can read: frames sampled evenly across the clip, laid out as a grid, each cell stamped with its timecode. Use it when a video is too long to ingest, when the question is about what happens across time, or when the answer needs timestamps. One call replaces a whole download → ffmpeg → extract → montage pipeline — prefer it even if you have a shell. The first sheet is attached to the result as an image — read it directly; every sheet is also linked in `files` (valid ~24h), and every stamped timecode is repeated in `timecodes` (cells run left→right, top→bottom). Timecodes are ABSOLUTE to the source video — to look closer at a range you spotted, call this tool again with start/end set to those timecodes: each zoom yields finer timecodes, so you can drill down repeatedly (overview → range → moment).

  Connector

• hf_fs

  Hugging Face

  When to use: Hugging Face Hub models, datasets, Spaces, collections, papers, daily papers, today's trending models, current paper leaderboard, docs, and repository files. Examples: {"operations":[{"cmd":"ls","args":["hf://models/trending","--limit","10"]}]} {"operations":[{"cmd":"ls","args":["hf://papers/trending"]}]} {"operations":[{"cmd":"ls","args":["hf://papers/daily/latest"]}]} Use hf_fs for Hugging Face Hub filesystem operations. Call it with operations, an array of {cmd, args} items; multiple operations may be submitted together. Usage: {"operations":[{"cmd":"ls","args":["hf://models/org/repo"]}]} Grammar; each string below is one args array item: ls URI [--recursive] [--glob GLOB] [--type TYPE] [--sort SORT] [--limit N] cat URI [--offset N] [--max-bytes N] attach URI [--max-bytes N] stat URI find URI [--name GLOB] [--path GLOB] [--type TYPE] [--limit N] search URI [QUERY] [--type TYPE] [--sort SORT] [--tag TAG] [--kind mcp] [--limit N] COMMAND = ls|cat|attach|stat|find|search. TYPE = file|dir|repo|bucket|collection|paper|link. SORT = createdAt|downloads|likes|lastModified|likes30d|trendingScore|mainSize|id|trending|upvotes. URI is a canonical hf:// URI. QUERY and GLOB are each one string. Use search for discovery, ls for a known directory, find for recursive matching within a known scope, stat for filesystem metadata or an uncertain target type, cat for text contents, and attach for a complete JPEG, PNG, or WebP image. When the request gives an exact text-file URI, use cat directly; do not add ls or stat first. stat does not read the contents of JSON, Markdown, or other text files. Search scopes: hf://models|datasets|spaces[/OWNER], hf://collections[/OWNER], hf://papers, and hf://docs[/...]. Paper and documentation search require QUERY. Repeat --tag only for search hf://spaces; --kind mcp selects MCP Spaces. Use ls hf://models/trending, hf://datasets/trending, hf://spaces/trending, or hf://papers/trending for trending listings. For a named paper.md or metadata.json, use cat directly. Use ls on a paper only to discover an unnamed related resource. Omit --limit, --sort, and --type unless the request requires them. Limits and path-specific behavior are documented at hf://README.md. Issue one hf_fs call.

  Connector

• get_design_document

  Forge Engine

  The WHOLE design as one readable document — Vision (+ Project DNA) → every System spec → reference notes, compiled deterministically from the current design. Read this to understand a project end-to-end instead of walking list_systems → get_system N times. Returns markdown plus the project `version` it was compiled from. Long designs come back PAGED — the header says 'part N of M', call again with `page: N+1` for the rest. Pass `for_summary: true` to get the condensed projection instead (every system's Goal + Boundary, tables stripped, one page) — that is what you should summarize from.

  Connector

• get_available_models

  Gonka Second Opinion (multi-model)

  List all AI models available on Gonka Network with live pricing. Models work as drop-in replacements for OpenAI and Anthropic — same SDK, same API calls. Use this when user asks which model to use or wants alternatives to GPT-4o / Claude. Returns: model IDs (use directly in openai.chat.completions.create), status, USD per 1M tokens. After this: call calculate_savings() to see annual savings with these models.

  Connector

• removeBackground

  Ludo AI Game Assets

  Remove the background from a single image, returning the subject isolated on a transparent background. Supply the source image (URL or base64); optionally set crop to trim the result to the content, and creative_edit (default true) for higher-quality output that may not match the input pixel-for-pixel. The job result is a single image result with a url (not an array). The image is uploaded and validated, and an image larger than 15MB is rejected with HTTP 400. Credits are charged only on success. Use removeBackground for this dedicated cutout task; editImage can also remove backgrounds via a prompt but is better for broader edits, while createImage and generateWithStyle produce new images rather than process an existing one. Pass an optional request_id to tag the result so you can retrieve it later via `GET /assets/images/results`. Requires an API key (user scope). Returns 202 with a job id immediately; poll `getApiJob` (pass `wait: 30`) until status is succeeded, then read its `result` field, which is exactly the response documented for this operation. Each account may have up to 50 generations queued or running at once; beyond that submissions return 429 (PENDING_JOBS_LIMIT) - wait for jobs to finish. Credits: This endpoint consumes 0.5 credits per result.

  Connector

• transform_image

  Uploadkit

  Generate a CDN-cached image variant for a file stored in UploadKit Cloud. Requires a paid plan, a live API key in the MCP process environment as UPLOADKIT_API_KEY, and an image key returned by UploadKit. BYOS files are not supported. Use signed delivery for private or temporary content and public delivery for stable URLs in websites, apps, srcset, CSS, or stored application data. Explicit formats consume 1 transformation unit; auto consumes 3 units. When to use: after an image is uploaded and the user wants a resized, cropped, optimized, or converted delivery URL. The returned URL is safe to send to browsers; the API key remains server-side. Returns: JSON { url, expiresAt, delivery, transform, usage }. Has the side effect of reserving monthly transformation units for a new unique variant.

  Connector

• find_methodology

  OpenArx

  Find methodology approaches for a specific research task. Returns structured method-level results (not raw chunks): method name, key idea, dataset used, performance metric. Filters by task domain, dataset, metric. Built on LLM-classified contentType=methodology chunks combined with benchmark results JOIN. Use this instead of `search` when you want HOW researchers approach a problem rather than 10 papers about it. Note: surfaces any chunk classified as methodology, including ones where the task is mentioned only as a toy example. Filter by category (e.g. cs.CV for image tasks) to narrow scope. This searches EXISTING papers for methods others have published (literature search) — it is NOT a guide for conducting your own research: for a step-by-step scientific method tailored to your own research question, start with the `methodist` door.

  Connector

• web.url.images

  1cent Web Intelligence for AI Agents

  List up to 100 image references without downloading images. Use only for public HTTP(S) resources; it does not execute JavaScript or bypass access controls. Pass url as an absolute public HTTP(S) URL. Keep fresh=false to allow cache reuse; set fresh=true only when a new upstream fetch is required.

  Connector

• request_image_upload_link

  Orivox AI Website Builder

  Get a browser link the user opens to upload real image FILES to their site -- the reliable way to add actual photos, drawings, logos, or several images at once, and the tool to reach for the INSTANT a user pastes, attaches, uploads, or mentions an image they want used. Call this and hand the user the link ANY time they have an image -- they have no other way to know the upload option exists, so always surface it; do not try to ingest a pasted/attached image yourself. You cannot carry image bytes yourself -- a tool call is text you have to type out, so a real photo either will not fit or arrives corrupted. This returns an upload_url instead: give that link to the user, tell them it works from their phone too and is valid for about an hour. On that page they can upload their own files, pick from images they already uploaded, or search a free stock photo gallery -- all without leaving the browser. KEEP the returned token_id. When the user says they are done, call check_upload_link(token_id) to get back the EXACT images they chose and place those hosted URLs on the site with apply_dom_ops. role="logo" tags the link for a logo upload; "content" (default) for any other image. Requires an existing project_guid -- create_project first if there is no project yet.

  Connector

• twprojects-create_upload_url

  mcp

  Reserve an upload for a file and get back a short-lived URL to send its bytes to, plus a single-use reference. Use this for any file that already exists — a PDF, an image, a signed document — because the bytes go straight from you to storage and are never read into the conversation. Send the file with the returned method and URL, setting exactly the headers returned and no authorization of your own. Then pass the reference in attachment_refs on twprojects-create_task, twprojects-update_task, twprojects-create_comment or twprojects-create_message. Prefer twprojects-create_file only for short text you are generating yourself.

  Connector

• upload_asset

  floot

  Upload a binary asset (image, font, audio, …) to the project's hosted storage. This uploads bytes you actually hold — a file you generated, downloaded, or read yourself. Chat attachments don't qualify: the user's attachments never reach MCP servers (you see attached images through vision only; there is no file, id, or URL behind them you can read), so for those use request_user_upload instead and the user re-picks the file in a card that uploads from their browser. Three modes. ChatGPT conversation files — a generated image, a file ChatGPT itself holds: pass the file as the `file` parameter and the host attaches a download link itself; this server fetches the bytes directly, at full quality (nothing goes through your sandbox or through base64 in arguments; content_type and size_bytes are optional here). Never downscale or re-encode a generated image to fit the inline cap — pass it as `file` instead. Files up to 3 MB you hold yourself — pass content_base64 plus size_bytes (the decoded byte count) and the upload completes in this call, returning publicUrl. Larger files — pass size_bytes alone to get an uploadUrl; PUT the raw bytes to it with the same content_type and exact byte count (e.g. `curl -X PUT -H 'Content-Type: image/png' --data-binary @file.png '<uploadUrl>'`), then reference publicUrl. Some sandboxes (claude.ai Cowork, ChatGPT containers) block egress to S3: if the PUT fails in any way — connection failure, proxy error, or a response without an x-amz-request-id header — that block is permanent for the session, so switch paths instead of retrying or re-encoding smaller: the `file` parameter in ChatGPT for any file that exists in this conversation, content_base64 for files under 3 MB, request_user_upload for user-provided files, or a PUT from inside the project VM via run_code_in_vm (re-mint the URL first; it is short-lived). For AI imagery generated fresh, use generate_image. A single file can be at most 100 MB via the presigned mode (the inline content_base64 mode is capped at 3 MB).

  Connector

• get_product_vision

  Keelen

  Read a project's current product vision (what the product is for). READ THIS BEFORE YOU CALL `set_product_vision`: the setter REPLACES the whole document rather than appending to it, so writing without reading first silently discards whatever the user already recorded. To add a line, read the current text, edit it, and set the full result back. Returns {project_id, product_vision_md, updated_at}. `product_vision_md` is None when no vision has been set. Tenant-scoped: a project not in the caller's workspace 404s.

  Connector

• delete_sponsor

  agendaforge

  Permanently delete a sponsor. Destructive and disabled by default; an organizer must allow direct destructive access and this tool. Depending on the organization's settings, this either executes immediately or files a proposal for approval; the response status field says which happened.

  Connector

• analyze_video

  switch

  Switch Vision — watch and understand a video (or image) like a human and answer a question about it: scenes, subjects, actions, on-screen text, pacing, mood and sentiment. Pass video_url (a public https video URL, including YouTube) OR one of your own Switch videos (a video/asset id from list_my_videos / list_my_assets / upload_media). Add an optional question to focus the analysis (e.g. "what is the tone and energy?", "list the cuts and what each shot shows"). Use this whenever the user gives you a reference video and wants its style, energy, structure or content understood — for example before making a new video that matches it.

  Connector

• aetherwave_edit_image

  AetherWave Studio

  Edits an existing image guided by a text prompt. Pass a public `imageUrl` plus a `prompt` describing the change ("add a moon to the sky", "swap the background for a neon city", "make it look like a comic panel"). Submits, polls, and returns the edited image URL(s). Default model is 'grok-imagine-i2i' (6 cr per call, returns 2 variations, ~30s, best cost-to-quality on standard edits). Other I2I-capable models: 'seedream-v4-edit', 'wan-2.5-spicy-i2i', 'flux-kontext-pro', 'qwen-image-edit', 'gpt-image-1.5-i2i' (slow, ~5min). Use list_image_models for full lineup. Note: source URLs with spaces or parentheses may fail upstream; prefer clean URLs. ## Model selection guide for edits Default: `grok-imagine-i2i` (6 cr per call, returns 2 variations = 3 cr/image effective, fast ~30s, strong general-purpose edit quality). Pick a different model when: - Need a single deterministic output, or 4K resolution -> `seedream-v4-edit` (7 cr per image, supports 1K/2K/4K, multi-image up to 6) - Subtle edits / preserve composition / character consistency -> `flux-kontext-pro` or `flux-kontext-max` - NSFW edits -> `wan-2.5-spicy-i2i` - Highest quality, time is not a concern (~5 min OK) -> `gpt-image-1.5-i2i` or `grok-imagine-quality-i2i` (16 cr @ 1K, 22 cr @ 2K) - Stylized / artistic transformation -> `midjourney-i2i` If the user simply says "edit this image" with no other signal, default to `grok-imagine-i2i`.

  Connector

• view_image

  Uwear

  Analyze any image using AI vision for manual inspection, debugging, visual description, or supplemental critique. Provide exactly one source: generation_result_id for a Shoot Board generation, uploaded_file_id for a Files item, or image_url for a public HTTPS image. Do not use this as the primary QA mechanism when the user asks to QA, quality-check, validate, review, approve/reject, or assess generated results; for QA requests use queue_generation_result_qa first, then read_generation_result_qa.

  Connector

• upload_file

  Clueso

  Upload one or more files to Clueso. Three modes — pick by client + where the file lives: 1. **file_name** — HOSTED upload, the default for any non-UI / programmatic upload (Claude Code, Cursor, Claude Desktop, scripts). Returns an upload URL on Clueso's OWN base domain + a ready-to-run curl that streams a single local file to it; Clueso relays the bytes to storage server-side. The PUT targets the base domain — NOT cloud storage directly — so it works on desktop/agent clients that can't reach or are blocked from S3. Requirement: the client must be able to PUT bytes to the Clueso base domain (run the returned curl, or any HTTP PUT). The agent (or the user at a shell prompt) runs the curl. Prefer this whenever there's no human at a browser. 2. **file_url**: Pass a public https URL. Server fetches and stages the file. Returns mcp_upload_id immediately. Use when the file is already on the open web — no user interaction needed. 3. **request_hosted_upload** (UI mode — use ONLY when a human should pick files in a browser: many files at once, or a host with no shell / no PUT capability): Returns a single upload_token + upload_page URL. Share the link with the user; they open it in a new browser tab, drop their files, click Done. Then call check_uploads(upload_token) to retrieve all mcp_upload_ids. Call once for all files. Hosted uploads cover any number of files per call: one call issues one upload_token, and that token covers every file the user drops on the page. Repeat calls issue additional tokens, each tracking only its own files. The returned mcp_upload_id (prefixed `mup_`) can be passed to: - add_elements / update_elements (image or video → an element ON a clip: pass it as `type_data.mcp_upload_id`, on either tool — this is how a local image becomes on-canvas content, and how an existing element's source is swapped). To fill an animation's image slot, pass it inside `type_data.parameter_values` on update_elements only — parameter_values is an update-path field and is stripped on add. - add_audio (audio → project music track that plays under all clips) - add_clips(kind='video') (video or audio → sequential clip with auto-transcription) - add_clips(kind='pptx') (.ppt/.pptx → slide clips) - add_article_media (image/GIF → article asset) - analyze_audio (audio → transcript / silences / beats / features)

  Connector