Model Card Generator

Documents

model-card-generator

Model Card Generator is an Agent Skill for documents published in the a5c-ai/babysitter repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Model documentation skill for generating model cards following Google's model card framework.

View on GitHub ↗

QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

Open your project in Codex.
Copy and paste the prompt below.
Review files and risks before approving any action.

Prompt to paste

Inspect the Agent Skill at https://github.com/a5c-ai/babysitter/blob/HEAD/library/specializations/data-science-ml/skills/model-card-generator/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/model-card-generator, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

model-card-generator

Overview

Model documentation skill for generating comprehensive model cards following Google's model card framework for ML model documentation.

Capabilities

Model details documentation (architecture, training, etc.)
Intended use specification
Performance metrics documentation
Ethical considerations section
Caveats and limitations
Quantitative analysis sections
Version history tracking
Multiple output formats (HTML, Markdown, JSON)

Target Processes

Model Interpretability and Explainability Analysis
Model Evaluation and Validation Framework
ML Model Retraining Pipeline

Tools and Libraries

Model Card Toolkit
TensorFlow Model Analysis (optional)
Jinja2 (templating)

Input Schema

{
  "type": "object",
  "required": ["modelDetails", "intendedUse"],
  "properties": {
    "modelDetails": {
      "type": "object",
      "properties": {
        "name": { "type": "string" },
        "version": { "type": "string" },
        "type": { "type": "string" },
        "architecture": { "type": "string" },
        "trainingDate": { "type": "string" },
        "framework": { "type": "string" },
        "citations": { "type": "array", "items": { "type": "string" } },
        "license": { "type": "string" }
      }
    },
    "intendedUse": {
      "type": "object",
      "properties": {
        "primaryUses": { "type": "array", "items": { "type": "string" } },
        "primaryUsers": { "type": "array", "items": { "type": "string" } },
        "outOfScopeUses": { "type": "array", "items": { "type": "string" } }
      }
    },
    "factors": {
      "type": "object",
      "properties": {
        "relevantFactors": { "type": "array", "items": { "type": "string" } },
        "evaluationFactors": { "type": "array", "items": { "type": "string" } }
      }
    },
    "metrics": {
      "type": "object",
      "properties": {
        "performanceMetrics": { "type": "array" },
        "decisionThresholds": { "type": "object" },
        "variationApproaches": { "type": "array" }
      }
    },
    "evaluationData": {
      "type": "object",
      "properties": {
        "datasets": { "type": "array" },
        "motivation": { "type": "string" },
        "preprocessing": { "type": "string" }
      }
    },
    "trainingData": {
      "type": "object",
      "properties": {
        "datasets": { "type": "array" },
        "motivation": { "type": "string" },
        "preprocessing": { "type": "string" }
      }
    },
    "ethicalConsiderations": {
      "type": "array",
      "items": {
        "type": "object",
        "properties": {
          "name": { "type": "string" },
          "mitigationStrategy": { "type": "string" }
        }
      }
    },
    "caveatsAndRecommendations": {
      "type": "array",
      "items": { "type": "string" }
    },
    "outputConfig": {
      "type": "object",
      "properties": {
        "format": { "type": "string", "enum": ["html", "markdown", "json"] },
        "outputPath": { "type": "string" }
      }
    }
  }
}

Output Schema

{
  "type": "object",
  "required": ["status", "modelCardPath"],
  "properties": {
    "status": {
      "type": "string",
      "enum": ["success", "error"]
    },
    "modelCardPath": {
      "type": "string"
    },
    "format": {
      "type": "string"
    },
    "sections": {
      "type": "array",
      "items": { "type": "string" }
    },
    "warnings": {
      "type": "array",
      "items": { "type": "string" },
      "description": "Warnings about missing recommended sections"
    }
  }
}

Usage Example

{
  kind: 'skill',
  title: 'Generate model card',
  skill: {
    name: 'model-card-generator',
    context: {
      modelDetails: {
        name: 'Fraud Detection Model',
        version: '2.0.0',
        type: 'Binary Classification',
        architecture: 'XGBoost',
        trainingDate: '2024-01-15',
        framework: 'scikit-learn',
        license: 'Proprietary'
      },
      intendedUse: {
        primaryUses: ['Transaction fraud detection'],
        primaryUsers: ['Risk management team'],
        outOfScopeUses: ['Credit scoring', 'Identity verification']
      },
      metrics: {
        performanceMetrics: [
          { name: 'AUC-ROC', value: 0.95 },
          { name: '[email protected]', value: 0.87 }
        ]
      },
      ethicalConsiderations: [
        { name: 'Demographic bias', mitigationStrategy: 'Regular fairness audits' }
      ],
      outputConfig: {
        format: 'markdown',
        outputPath: 'docs/model_card.md'
      }
    }
  }
}