More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Rasterless: PNG to SVG with AI

    3

    PNG to SVG with AI. Turns icons, logos, and flat art into small, clean SVG files.

    Free

    4 usesSecurity scanned
    Rasterless: PNG to SVG with AI

    Rasterless: PNG to SVG with AI

    Example session with this skill installed

    Convert flying-saucer.png and the icons in ./icons to clean SVGs for our website.

    • Read your context and instructions
    • Compiled the rasterless

    I converted 3 images with rasterless and checked each one against its original.

    FileColorsSizeMismatch
    flying-saucer.svg6 colors12.7 KB0.00%
    icons/house-fill.svg1 color, currentColor0.7 KB0.00%
    icons/bell-fill.svg1 color, currentColor1.1 KB0.00%

    The script's report for the main file

    flying-saucer.svg: 6 colors [#ffd983 #ccd6dd #9aaab4 #5dadec #67757f #8ccaf7], 554 segments, 12.7 KB, mismatch 0.00%
    

    I opened each preview (original, SVG render, misses in red). All three are clean. The saucer's six colors came through, including the small light-blue lights on the rim, and no misses are marked.

    Next steps

    • The two icons use fill="currentColor", so color: #FF6B4A in your CSS recolors them.
    • Need exact hex colors for Figma? Ask for design mode and I will re-run with --design.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Rasterless turns PNG, JPG, and WebP icons, logos, and flat illustrations into small, layered SVG files that look hand-drawn, and checks every result against the original.

    What it does

    • One path per color. A 6-color emoji becomes 6 clean layers.
    • True colors. The palette comes from the flat insides of shapes, so anti-aliased edges never become fake colors.
    • Sub-pixel edges. The image is upscaled and snapped before tracing, so curves land between pixels.
    • No hairline gaps. Layers overlap slightly, so the background never shows through a seam.
    • Self-check. Each SVG is drawn back and compared with the input. Misses are counted and marked in red on a preview.
    • Shaded mode. Grayscale 3D-style art becomes smooth stacked gray bands, with glossy balls rebuilt as real radial gradients.

    Results

    Tested on 384 images the converter was not tuned on: 100% pass on 512 px PNGs, 99% on JPGs, and files about a third the size of VTracer's output.

    Good to know

    • Built for flat art. Photos and color gradients are not a good fit, and the converter tells you when it sees them.
    • Single-color icons use currentColor, so CSS can recolor them.
    • Free and open source (MIT). Source, samples, and benchmarks are on GitHub: dheerajponnaganti/rasterless

    Not for photos. Color gradients come out as flat bands. Detailed icons under 64 px lose thin lines. Shaded mode handles grayscale art only.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    4 uses

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access

    Runs its Python scripts in the terminal, reads the image you give it, and writes the SVG next to it. The first run downloads Pillow and NumPy from PyPI. Your images are never uploaded anywhere.

    Listed1 month ago
    Updated1 month ago

    What's inside

    Frequently Asked Questions

    It is built for flat art. It reads colors from the flat insides of shapes, traces one path per color, and checks the result against the original. On a 384-image test it passed 100% of 512 px PNGs, with files about a third the size of VTracer's.

    Icons, logos, emoji, and flat illustrations as PNG, JPG, or WebP, ideally 256 px or larger. Grayscale 3D-style art has its own shaded mode. Photos and color gradients are not a good fit.

    Every run draws the SVG back and compares it with the input. It reports the share of visibly different pixels and can save a preview with any misses marked in red.

    It is free. You get the skill folder: SKILL.md, the two Python scripts, and a bundled copy of the Potrace tracer. It is MIT licensed, and the source is on GitHub.

    It follows the open SKILL.md standard. It was tested in Claude Code, Codex, Claude.ai, and ChatGPT (Work mode). It needs Python 3.10 or newer where it runs.

    Yes. Single-color icons use currentColor, so Tailwind text classes or CSS color recolor them. In JSX, import the file with SVGR or use it as an img. If you paste the markup inline, rename fill-rule to fillRule.

    New in Image & Video