More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Tracefield

    1

    Turn a photograph into an animated, editable constellation overlay: labelled anchor points on your subject, a decorative node-link web and an animated trace, edited and exported from a local studio.

    Free

    Security scanned
    Tracefield

    Tracefield

    Example session with this skill installed

    Here's a photo of a red-eyed tree frog on moss: /Users/me/Pictures/tree-frog.png. Make a Tracefield constellation study of the frog and open it in the studio.

    • Read your context and instructions
    • Compiled the tracefield

    Summary

    Created study tree-frog and opened it in the local studio. The web follows the frog's eye, snout, back and limbs; the moss and bokeh are deliberately left out.

    This example is the Tree Frog study bundled in Tracefield's public gallery, so you can open the same scene yourself.

    Roles declared

    RoleVisible criterionAnchors
    eyeVisible red eye with dark pupil1
    snoutVisible nose or mouth edge2
    backVisible upper back or side boundary4
    jointVisible bend or junction of a limb6
    toeVisible orange toe pad or foot extremity7

    20 observation points · 28 decorative links · focus: eye.01

    Sample anchors

    LabelWhy it's there
    eye.01Visible red eye and pupil
    snout.01Small visible nostril
    back.03Back behind the head
    joint.04Front limb shoulder
    toe.03Lower hind toe pad

    How the skill checks its work

    • Rebuilds the web around the authored points (npm run net) and validates the scene (npm run validate)
    • Opens the studio on a local loopback port, then reviews the settled frame and one full trace cycle: labels should sit on visible details and no link should reach into the background

    Next steps

    • In the studio, try a different Blend mode or Marker style
    • Use

    Export → Image for a PNG,

    Loop for an MP4/WebM, or

    Project to save a portable file you can reopen later without re-interpreting the photo

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

    About this skill

    Give your agent a photo and Tracefield turns it into a subject-bound constellation study: labelled anchor points on the visible details of your subject, a decorative node-link web that hugs its silhouette, and an animated trace that draws it in.

    How it works

    The agent inspects the image, declares 4–7 visible roles (an eye, a petal tip, a rim) and places roughly 12–24 points, each with a short reason tied to a visible detail. It then opens a local studio where you adjust the look yourself.

    What you get

    • Blend modes and marker styles, edited live in the browser
    • Export a PNG, a looping MP4/WebM, or a portable project you can reopen later
    • The studio and rendering run on your machine with no model calls; studies stay local and out of Git (your agent's own image inspection follows its usual provider settings)
    • Saved scenes are reproducible: the same scene gives the same web and timeline

    Good for HUD/FUI-style treatments, poster studies, motion pieces and social visuals.

    Requirements

    git, Node 24+, npm and a browser. On first run the bundled setup script fetches the pinned v0.1.0 release from GitHub and installs its locked dependencies.

    Source

    Open source (MIT): https://github.com/rishi222k/tracefield

    Needs git, Node 24+, npm and a browser. PNG/JPEG up to 20 MB and 32 MP. JPEGs that rely on EXIF rotation must first be exported as upright PNGs. The web is a decorative aesthetic, not object detection; a fresh interpretation of the same photo can differ (saved scenes are what make results repeatable). Loop export format (MP4 or WebM) depends on your browser's encoders.

    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

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 21 days ago

    • Free with an Agensi account

    Requires

    git
    Node.js 24+
    npm
    Web browser

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Browser

    File Scopes

    ~/tracefield-lab/**

    Terminal: runs the bundled setup script and the lab's npm commands. Network: setup clones the pinned v0.1.0 release from github.com/rishi222k/tracefield and runs npm ci for its locked dependencies; nothing else is fetched and no images are uploaded. Read/write files: works only inside the lab directory (~/tracefield-lab by default, or $TRACEFIELD_LAB / --lab), where it copies your photo and writes the scene. Browser: opens the local studio on a loopback port (4318-4418) to check the result.

    Listed21 days ago

    What's inside

    Frequently Asked Questions

    Your agent looks at the photo, declares 4–7 visible roles (like eye or petal tip) and places each point on a visible detail with a short reason. Tracefield then draws a decorative web between those points that stays close to the subject. It's an aesthetic layer, not object detection, so you can review and adjust it.

    Yes. Tracefield is open source under the MIT licence, and the full source is on GitHub at github.com/rishi222k/tracefield.

    Any agent that can view an image, run shell commands and open a browser, such as Claude Code or Codex CLI. The agent does the interpretation; Tracefield's local studio handles rendering and export.

    Tracefield itself makes no model calls and keeps studies in a local folder that's ignored by Git. Your agent viewing the photo follows that agent's own provider and privacy settings.

    A PNG image, a looping MP4 or WebM video (depending on your browser's encoders), or a portable project file that restores the image, scene and settings without re-interpreting the photo.

    New in Image & Video