FedMCP Python Core

Python implementation of the Federal Model Context Protocol (FedMCP) v0.2.

Installation

pip install fedmcp

For development:

pip install -e ".[dev]"

Quick Start

Creating and Signing Artifacts

from uuid import uuid4
from fedmcp import Artifact, ArtifactType, LocalSigner

# Create an artifact
artifact = Artifact(
    type=ArtifactType.AGENT_RECIPE,
    workspaceId=uuid4(),
    jsonBody={
        "name": "Healthcare Diagnostic Agent",
        "version": "1.0.0",
        "capabilities": ["diagnose", "recommend", "monitor"]
    }
)

# Sign it
signer = LocalSigner()
jws_token = signer.sign(artifact)
print(f"Signed artifact: {jws_token[:50]}...")

Verifying Artifacts

from fedmcp import Verifier

# Create verifier and add public key
verifier = Verifier()
verifier.add_public_key(signer.get_key_id(), signer.private_key.public_key())

# Verify the artifact
verified_artifact = verifier.verify(jws_token)
print(f"Verified artifact ID: {verified_artifact.id}")

Using the Client

from fedmcp import FedMCPClient
import asyncio

async def main():
    client = FedMCPClient(
        base_url="http://localhost:8000",
        workspace_id=uuid4(),
        signer=LocalSigner()
    )
    
    # Create an artifact
    result = await client.create_artifact(
        artifact_type=ArtifactType.LLM_COMPLETION,
        json_body={
            "model": "gpt-4",
            "prompt": "Diagnose patient symptoms",
            "completion": "Based on the symptoms..."
        }
    )
    
    print(f"Created artifact: {result['id']}")
    client.close()

asyncio.run(main())

AWS KMS Integration

from fedmcp import KMSSigner, KMSVerifier

# Sign with KMS
kms_signer = KMSSigner(
    kms_key_id="arn:aws:kms:us-gov-west-1:123456789012:key/abc-123",
    region="us-gov-west-1"
)
jws_token = kms_signer.sign(artifact)

# Verify with KMS
kms_verifier = KMSVerifier(region="us-gov-west-1")
kms_verifier.add_kms_key(
    key_id=kms_signer.get_key_id(),
    kms_key_id="arn:aws:kms:us-gov-west-1:123456789012:key/abc-123"
)
verified = kms_verifier.verify(jws_token)

Artifact Types

Standard FedMCP artifact types:

• SSP_FRAGMENT - System Security Plan fragments
• POAM_TEMPLATE - Plan of Action & Milestones templates
• AGENT_RECIPE - AI agent configurations
• BASELINE_MODULE - Security baseline modules
• AUDIT_SCRIPT - Audit automation scripts

Healthcare-specific types:

• RAG_QUERY - Retrieval-augmented generation queries
• LLM_COMPLETION - Language model completions
• TOOL_INVOCATION - Tool/function call records

Testing

Run tests with pytest:

pytest

License

Apache 2.0 - See LICENSE file for details.