Skip to main content

Cryptographic schema integrity verification for AI tools

Project description

SchemaPin Python Implementation

A Python implementation of the SchemaPin protocol for cryptographic schema integrity verification of AI tools.

Overview

SchemaPin provides cryptographic verification of AI tool schemas using ECDSA P-256 signatures and Trust-On-First-Use (TOFU) key pinning. This Python implementation serves as the reference implementation for the protocol.

Features

  • ECDSA P-256 Cryptography: Industry-standard elliptic curve signatures
  • Schema Canonicalization: Deterministic JSON serialization for consistent hashing
  • Public Key Discovery: Automatic retrieval from .well-known/schemapin.json endpoints
  • Key Pinning: Trust-On-First-Use security model with SQLite storage
  • High-Level Workflows: Simple APIs for both developers and clients
  • Comprehensive Testing: Full test suite with security validation

Installation

From PyPI (Recommended)

# Install latest stable version
pip install schemapin

# Install with development dependencies
pip install schemapin[dev]

# Install with testing dependencies only
pip install schemapin[test]

From Source (Development)

# Clone repository and install in development mode
git clone https://github.com/thirdkey/schemapin.git
cd schemapin/python
pip install -e .[dev]

After installation, the following CLI tools will be available:

  • schemapin-keygen - Generate cryptographic key pairs
  • schemapin-sign - Sign JSON schemas
  • schemapin-verify - Verify signed schemas

CLI Tools

SchemaPin provides three command-line tools for common operations:

Key Generation (schemapin-keygen)

Generate ECDSA or RSA key pairs with optional .well-known template:

# Generate ECDSA key pair with .well-known template
schemapin-keygen --type ecdsa --developer "Your Company" --well-known

# Generate RSA 4096-bit key pair
schemapin-keygen --type rsa --key-size 4096 --output-dir ./keys

# Generate keys in DER format
schemapin-keygen --type ecdsa --format der --prefix mykeys

Schema Signing (schemapin-sign)

Sign JSON schema files with private keys:

# Sign a single schema
schemapin-sign --key private.pem --schema schema.json --output signed.json

# Sign with metadata
schemapin-sign --key private.pem --schema schema.json --developer "Your Company" --version "1.0"

# Batch sign multiple schemas
schemapin-sign --key private.pem --batch ./schemas/ --output-dir ./signed/

# Sign from stdin
echo '{"type": "object"}' | schemapin-sign --key private.pem --stdin

Schema Verification (schemapin-verify)

Verify signed schemas with public keys or discovery:

# Verify with public key
schemapin-verify --schema signed.json --public-key public.pem

# Verify with domain discovery and interactive pinning
schemapin-verify --schema signed.json --domain example.com --tool-id my-tool --interactive

# Batch verify with auto-pinning
schemapin-verify --batch ./signed/ --domain example.com --auto-pin

# Verify from stdin with JSON output
echo '{"schema": {...}, "signature": "..."}' | schemapin-verify --stdin --public-key public.pem --json

CLI Examples

Run the CLI examples script to see detailed usage patterns:

cd python/examples
python cli_usage_examples.py

Quick Start

Tool Developer Workflow

from schemapin.utils import SchemaSigningWorkflow, create_well_known_response
from schemapin.crypto import KeyManager

# 1. Generate key pair
private_key, public_key = KeyManager.generate_keypair()
private_key_pem = KeyManager.export_private_key_pem(private_key)
public_key_pem = KeyManager.export_public_key_pem(public_key)

# 2. Sign your tool schema
schema = {
    "name": "calculate_sum",
    "description": "Calculates the sum of two numbers",
    "parameters": {
        "type": "object",
        "properties": {
            "a": {"type": "number", "description": "First number"},
            "b": {"type": "number", "description": "Second number"}
        },
        "required": ["a", "b"]
    }
}

signing_workflow = SchemaSigningWorkflow(private_key_pem)
signature = signing_workflow.sign_schema(schema)

# 3. Create .well-known response
well_known_response = create_well_known_response(
    public_key_pem,
    "Your Organization",
    "contact@yourorg.com"
)

# Host well_known_response at https://yourdomain.com/.well-known/schemapin.json

Client Verification Workflow

from schemapin.utils import SchemaVerificationWorkflow

verification_workflow = SchemaVerificationWorkflow()

# Verify schema with automatic key pinning
result = verification_workflow.verify_schema(
    schema,
    signature,
    "yourdomain.com/calculate_sum",
    "yourdomain.com",
    auto_pin=True
)

if result['valid']:
    print("✅ Schema signature is valid")
    if result['first_use']:
        print("🔑 Key pinned for future use")
else:
    print("❌ Schema signature is invalid")
    print("Error:", result['error'])

API Reference

Core Classes

SchemaPinCore

  • canonicalize_schema(schema) - Convert schema to canonical string format
  • hash_canonical(canonical) - SHA-256 hash of canonical string
  • canonicalize_and_hash(schema) - Combined canonicalization and hashing

KeyManager

  • generate_keypair() - Generate new ECDSA P-256 key pair
  • export_private_key_pem(private_key) - Export private key to PEM format
  • export_public_key_pem(public_key) - Export public key to PEM format
  • load_private_key_pem(pem_data) - Load private key from PEM
  • load_public_key_pem(pem_data) - Load public key from PEM

SignatureManager

  • sign_hash(hash_bytes, private_key) - Sign hash with private key
  • verify_signature(hash_bytes, signature, public_key) - Verify signature
  • sign_schema_hash(schema_hash, private_key) - Sign schema hash
  • verify_schema_signature(schema_hash, signature, public_key) - Verify schema signature

PublicKeyDiscovery

  • fetch_well_known(domain) - Fetch .well-known/schemapin.json
  • get_public_key_pem(domain) - Get public key from domain
  • get_developer_info(domain) - Get developer information

KeyPinning

  • pin_key(tool_id, public_key_pem, domain, developer_name) - Pin public key
  • get_pinned_key(tool_id) - Get pinned key for tool
  • is_key_pinned(tool_id) - Check if key is pinned
  • list_pinned_keys() - List all pinned keys
  • remove_pinned_key(tool_id) - Remove pinned key

High-Level Workflows

SchemaSigningWorkflow

workflow = SchemaSigningWorkflow(private_key_pem)
signature = workflow.sign_schema(schema)

SchemaVerificationWorkflow

workflow = SchemaVerificationWorkflow()
result = workflow.verify_schema(schema, signature, tool_id, domain, auto_pin)

Examples

Run the included examples:

# Tool developer workflow
cd python/examples
python tool_developer.py

# Client verification workflow  
python client_verification.py

Testing

cd python
python -m pytest tests/ -v

# Run code quality checks
ruff check .
bandit -r . --exclude tests/

Requirements

  • Python 3.8 or higher
  • cryptography library for ECDSA operations
  • requests library for HTTP operations
  • sqlite3 (built-in) for key storage

Security Considerations

  • Private Key Security: Store private keys securely and never expose them
  • HTTPS Required: Always use HTTPS for .well-known endpoint discovery
  • Key Pinning: Review pinned keys periodically and verify authenticity
  • Signature Verification: Always verify signatures before using tool schemas

Cross-Language Compatibility

This Python implementation is designed to be fully compatible with the JavaScript implementation:

  • Identical schema canonicalization results
  • Compatible ECDSA P-256 signatures
  • Same .well-known endpoint format
  • Interoperable key formats (PEM)

License

MIT License - see LICENSE file for details.

Contributing

  1. Fork the repository
  2. Create a feature branch
  3. Add tests for new functionality
  4. Ensure all tests pass
  5. Submit a pull request

Support

For issues and questions:

  • GitHub Issues: SchemaPin Repository
  • Documentation: See TECHNICAL_SPECIFICATION.md
  • Examples: Check the examples/ directory

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

schemapin-1.3.0.tar.gz (58.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

schemapin-1.3.0-py3-none-any.whl (36.5 kB view details)

Uploaded Python 3

File details

Details for the file schemapin-1.3.0.tar.gz.

File metadata

  • Download URL: schemapin-1.3.0.tar.gz
  • Upload date:
  • Size: 58.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.14

File hashes

Hashes for schemapin-1.3.0.tar.gz
Algorithm Hash digest
SHA256 fbce160bce53ea9931f007accae80877aeca89a27ffb36ae64c0979e65c93b38
MD5 a4fee678cc753aa09a52050d1e789a74
BLAKE2b-256 65d10af258d660e064f199b0bfc850d283c86ec12888b79a4fbb1e290e570a95

See more details on using hashes here.

File details

Details for the file schemapin-1.3.0-py3-none-any.whl.

File metadata

  • Download URL: schemapin-1.3.0-py3-none-any.whl
  • Upload date:
  • Size: 36.5 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.14

File hashes

Hashes for schemapin-1.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 1221757683a9dac688a0cb0107717b7fbb035873bd8c5878bc1449745a04aae6
MD5 c3447625b70695b4b1963c7023459509
BLAKE2b-256 1125f7c17171cc12293072f133ba9cc6245956621abc6a960c353f3a01ac212c

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page