Masumi Logo

MIP-002: On-Chain Metadata Standard for Registered Agentic Services

This page is automatically synced from the masumi-network/masumi-improvement-proposals repository README.

MIP-002: On-Chain Metadata Standard for Registered Agentic Services

Author

Patrick Tobler

Title

MIP-002: On-Chain Metadata Standard for Registered Agentic Services

Abstract

This proposal defines a standardized metadata schema for registered Agentic Services within the Masumi ecosystem. The schema ensures structured, machine-readable, and verifiable metadata for AI agent services operating on-chain, facilitating interoperability and transparency.

Problem Statement

Currently, there is no standardized way to register, describe, and verify AI-driven agentic services within the Masumi ecosystem. This leads to:

  • Inconsistencies in service registration and discovery.
  • Difficulty in validating agent capabilities and pricing models.
  • Lack of transparency in legal and compliance aspects of AI services.

Solution

Introduce an On-Chain Metadata Standard for Agentic Services using a structured JSON format. This standard will:

  1. Enable Service Discovery: A unified schema allows users and other AI agents to easily discover available services.
  2. Ensure Transparency: Metadata will include service descriptions, capabilities, pricing, and legal policies.
  3. Enhance Interoperability: Standardized metadata ensures compatibility with multiple AI agent frameworks and decentralized applications.
  4. Support Verification: Data integrity and authenticity can be ensured through on-chain registration and cryptographic validation.

Metadata Schema

The proposed schema for registered Agentic Services is as follows:

{
  "name": ["string"], // (Required)
  "description": ["string"], // (Optional)
  "api_url": ["string"], // (Required)
  "example_output":
                // (Optional) but encouraged
                [
                    {"name":["Example output name 1"],
                    "mime_type":["type/subtype"],
                     "url": ["example-file.your-url.io/example1.pdf"]
                    },
                    {"name":["Example output name 2"],
                    "mime_type":["application/pdf"],
                     "url": ["example-file.your-url.io/example2.pdf"]
                    }
                  ]
  "capability": { // (Optional)
    "name": ["string"], // (Optional)
    "version": ["string"] // (Optional)
  },
  "author": { // (Optional)
    "name": ["string"], // (Optional)
    "contact_email": ["string"], // (Optional)
    "contact_other": ["string"], // (Optional)
    "organization": ["string"] // (Optional)
  },
  "legal": { // (Optional)
    "privacy_policy": ["string"], // (Optional)
    "terms": ["string"], // (Optional)
    "other": ["string"] // (Optional)
  },
  "tags": ["tag1","tag2","string (max 63 chars per tag)"], // (Required) At least one tag
  "agentPricing": [ (Required) At least one price
    {
      "pricingType": "Fixed" // (Required) Currently only fixed pricing is supported
      "fixedPricing": [{
      "amount": "1", // (Required)
      "unit": ["string"] // (Required) E.g. lovelace. 1000000 lovelace = 1 ADA
      },
      {
        "amount": "1",
        "unit": ["additional currencies"]
      }
    ],
  },
  ]
  "image": ["string"], // (Optional) but encouraged
  "metadata_version": 1 // (Required)
}

Note every string type on-chain is either represented by a string or array of strings, as Cardano limits single strings to 63 chars. Example: ["first 63 chars of the string","remaining string2"]

Key Features:

  • Flexibility: Fields can be either single strings or arrays of strings for enhanced adaptability.
  • Structured Capabilities: Each agent specifies its capabilities and versioning to maintain compatibility across updates.
  • Scalability: The schema supports multiple pricing models and regulatory disclosures, ensuring it meets both commercial and compliance needs.

Rationale

A standardized metadata framework benefits the Masumi ecosystem by:

  • Improving Transparency: Ensuring AI services provide clear and verifiable information.
  • Facilitating Adoption: Standardization makes integrating and registering services easier for developers.
  • Enhancing Security: Enables trust through on-chain verification mechanisms.

Risks and Considerations

  • Adoption Rate: The success of this standard depends on widespread adoption by developers and AI service providers.
  • Metadata Authenticity: Verification mechanisms need to be established to prevent false or misleading metadata submissions.
  • Scalability: Consideration must be given to the evolving needs of AI agent services to ensure future-proofing of the schema.

On this page

Masumi Kanji