Skip to content
OpenAgentSkill
Browsing is open · Catalog captured 2026-11-21 · Automatic updates and account services are paused.
Design & UI

eye-art-polyphemus

Create and refine images with Eye.Art Polyphemus through its public MCP, including artist-guided images, reference edits, web visuals, editable SVG, and supported motion.

jacobwell · Static checks on file

Installation

Read the source and dependencies before running a command. Copying does not install or execute anything.

npx skills add jacobwell/skills --skill eye-art-polyphemus

About this skill


name: eye-art-polyphemus description: Create and refine images with Eye.Art Polyphemus through its public MCP, including artist-guided images, reference edits, web visuals, editable SVG, and supported motion.

Eye.Art Polyphemus

Use Eye.Art when the user wants an image or editable vector and the current agent host can connect to remote Streamable HTTP MCP servers.

Connect

Add this remote MCP server to the agent host:

URL: https://eye.art/api/eye-mcp Transport: Streamable HTTP Authentication: none

Use its eye_art_* tools only when they are available in the current host. If they are not available, explain that the endpoint must first be added as an MCP server. Do not imply a generation occurred without a completed tool result.

Choose a route

  • For a new image, use eye_art_make_image with mode: "image".
  • For a visual muse, pass artist: polyphemus, dali, goya, matisse, leonardo, van_gogh, rothko, or ross. Preserve the user's subject and composition; treat artist names as visual direction, not endorsement or exact imitation.
  • For icons and compact illustrations, use mode: "small_art" and specify the target size, silhouette, contrast, and background.
  • For an image matched to a website, use mode: "site_match", attach relevant HTML/CSS/JS/TS in pageReferences, and specify the target section, crop, palette, and clear space for copy.
  • For an image change, use eye_art_edit_image with the source image as imageDataUrl. Name exactly what should change and what should remain. Keep the source attached on follow-up edits if needed.
  • For a genuinely editable vector, use eye_art_make_svg; save the returned svg field as an .svg file. This makes vector geometry; it does not trace raster references or animate SVG.
  • Use eye_art_prompt_ideas when the user wants to explore directions before rendering. Use mode: "motion" only for supported motion requests.

Keep the conversation and retrieve results

  1. For related turns, pass the returned conversationId to the next make/edit call. Start a fresh conversation or set clearReferences: true for an unrelated concept.
  2. If a make/edit call returns a jobId, poll eye_art_image_status until the job is complete or failed. Show the result only after completion.
  3. For an edit, inspect the returned image against the source. If it drifts, retry with the source image again and a narrower change instruction.

Limits and privacy

  • No Eye.Art key is needed. Anonymous image generation is currently limited to 20 generations per hour per caller network identity; free to try does not mean unlimited.
  • A longer staged workflow can take several minutes. Relay the actual queued/running status and wait for its result; do not promise a fixed completion time.
  • Image references are sent to Eye.Art. Do not upload private or sensitive images without the user's direction. Images, prompts, and conversations may be retained for up to 30 days for service history and debugging.
  • Report actual tool status and failures. Do not claim provider, workflow, speed, or quality guarantees the result does not establish.

Tool reference

The live MCP endpoint exposes eye_art_make_image, eye_art_make_svg, eye_art_edit_image, eye_art_prompt_ideas, eye_art_image_status, and conversation list/get/delete tools. Check the endpoint's current tool schema for argument details:

https://eye.art/api/eye-mcp