More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Sign Language Translator (ASL A–Y)

    1

    CLI-based ASL alphabet translator using MediaPipe and TensorFlow Lite to convert hand gesture images into JSON.

    Free

    Security scanned

    Sign Language Translator (ASL A–Y)

    Example session with this skill installed

    Analyze the gesture in hand_sign_test.jpg and tell me what letter it is.

    • Read your context and instructions
    • Compiled the sign language translator

    I've analyzed the image using the HandLandmarker pipeline. The model successfully detected a hand and identified the sign.

    Translation Results

    FieldValue
    Detected LetterB
    Confidence Score0.9845
    Hand DetectedTrue
    OrientationOriginal
    StatusOK

    The hand gesture clearly represents the letter

    B. The high confidence score of 98.45% indicates a very reliable detection with good lighting and positioning.

    Next steps

    • If you have more images, provide the file path to translate them.
    • If the output needs to be voiced, I can send this letter to the Text-to-Speech tool.
    • For letters J or Z, please note that static images are not supported.

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

    About this skill

    Processing hand gestures for translation is often slow and inaccurate without specialized vision pipelines. This skill provides a lightweight CLI bridge that translates static images of American Sign Language (ASL) into structured JSON data.

    What it does

    • Alphabet translation classifies 24 static letters from the ASL dactylic alphabet (A-Y).
    • Hand landmarking detects 21 specific 3D points on the hand to ensure anatomical accuracy.
    • Confidence scoring returns a decimal value representing the model certainty for every detection.
    • Failure handling identifies when no hand is present or when image quality prevents a reading.
    • Environment management handles its own Python virtual environment and dependency installation automatically.

    How it works

    1. Provide image the user or agent passes a local path to a .jpg, .png, or .webp file.
    2. Execute command the agent runs run.py --imagen <PATH> to trigger the MediaPipe and TensorFlow Lite pipeline.
    3. Parse output the skill returns a standardized JSON object to stdout containing the prediction and confidence.
    4. Respond to user the agent delivers a concise translation or requests a clearer photo if confidence falls below 0.75.

    Frameworks & tools

    Built with Python 3.12, utilizing MediaPipe HandLandmarker for point detection and TensorFlow Lite for lightweight inference.

    Why this beats prompting it yourself

    Standard LLM vision models often hallucinate hand positions or confuse similar letters like M, N, and T. This skill uses a dedicated 21-point hand landmarking model to provide precise spatial analysis before classification, ensuring higher reliability than generic visual descriptions.

    Use cases

    • Accessibility tools convert finger-spelling photos into text for non-signers.
    • Educational feedback verify if a student is forming a specific ASL letter correctly.
    • Dataset labeling automate the initial tagging of hand gesture image libraries.

    Known limitations

    Does not support J or Z as they require motion tracking. Requires clear lighting and a visible hand to function.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Does not support J or Z letters due to motion requirements. Requires Python 3.12 and good lighting. Static images only.

    Open docs

    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 buying a skill.

    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 to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed today

    • Free with an Agensi account

    Software
    Git

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Environment Variables

    Allowed Hosts

    storage.googleapis.com
    Listedtoday

    What's inside

    Frequently Asked Questions

    This skill focuses specifically on the 24 static letters of the ASL alphabet (A-Y). It does not currently support 'J' or 'Z' because those letters require motion-tracking for their dynamic movements, which is outside the scope of this static image translator.

    Unlike general-purpose vision models that might guess based on the overall look of a hand, this skill uses a specialized 21-point landmarking system. This ensures the AI verifies the exact anatomical position of every finger, leading to significantly higher accuracy for similar-looking signs like 'M' and 'N'.

    As a CLI-based Python tool, it is compatible with any AI agent or framework that can execute shell commands and parse JSON. It is ideal for developers using AutoGPT, LangChain, or custom-built agents that need a reliable bridge to local computer vision processing.

    You receive the full source code for the CLI bridge, the pre-trained TensorFlow Lite model, and an automated setup script. The script handles all Python dependencies and environment configuration to ensure the skill is ready to run immediately after download.

    For the best results, the image should be well-lit with the hand clearly visible and facing the camera. If the lighting is too dim or the hand is obscured, the skill will return a 'failure' status in the JSON output rather than providing a low-confidence guess.

    New in Agents & Orchestration

    Free