nanda-model-provenance

Model provenance metadata for NANDA-compatible agent discovery.

A single dataclass that serializes model identity and versioning into the JSON shapes expected by NANDA AgentFacts, AgentCard, and decision-envelope outputs. Zero runtime dependencies.

Installation

pip install nanda-model-provenance

Quick Start

from nanda_model_provenance import ModelProvenance

provenance = ModelProvenance(
    model_id="llama-3.1-8b",
    model_version="1.0.0",
    provider_id="ollama",
    model_type="base",
    governance_tier="standard",
)

provenance.to_dict()                    # {"model_id": "llama-3.1-8b", ...}
provenance.to_agentfacts_extension()    # {"x_model_provenance": {...}}
provenance.to_agent_card_metadata()     # {"model_info": {...}}
provenance.to_decision_fields()         # {"model_id": ..., "provider_id": ...}

Output Formats

ModelProvenance
    ├── to_agentfacts_extension()  →  {"x_model_provenance": {...}}
    ├── to_agent_card_metadata()   →  {"model_info": {...}}
    └── to_decision_fields()       →  {"model_id": ..., "provider_id": ...}

Format	Target	Fields included
AgentFacts extension	NANDA AgentFacts metadata	All non-empty fields under x_model_provenance
AgentCard metadata	NANDA AgentCard model_info	All non-empty fields under model_info
Decision envelope	Flat top-level fields	model_id, model_version, provider_id only

The default AgentFacts extension key is x_model_provenance (vendor-neutral, following the NANDA x_ prefix convention). Vendors can use their own namespace:

provenance.to_agentfacts_extension(extension_key="x_myvendor")

Model Types

Type	Description
base	Foundation / pre-trained models
lora_adapter	LLM LoRA fine-tuned adapters
onnx_edge	Edge ONNX / TFLite models
federated	Federated learning models
heuristic	Rule-based / heuristic fallbacks

Governance Tiers

Tier	Description
standard	Default governance classification
regulated	Subject to additional policy requirements

Risk Levels

Level	Description
low	Minimal blast radius
medium	Moderate impact on failure
high	Requires additional governance approval

API

ModelProvenance

Field	Type	Default	Description
model_id	str	(required)	Model identifier (e.g. "llama-3.1-8b")
model_version	str	""	Semantic or arbitrary version string
provider_id	str	""	Inference provider ("openai", "ollama", "local")
model_type	str	""	Model category (see Model Types)
base_model	str	""	Foundation model name (when model_type is an adapter)
governance_tier	str	""	Governance classification (see Governance Tiers)
weights_hash	str	""	SHA-256 hex digest of model weights
risk_level	str	""	Risk assessment (see Risk Levels)

All optional fields default to "" and are omitted from serialized output.

Methods: to_dict(), to_agentfacts_extension(), to_agent_card_metadata(), to_decision_fields(), from_dict()

Related Packages

Package	Question it answers
nanda-model-provenance (this package)	"Where did this model come from?" (identity, versioning, provider, NANDA serialization)
nanda-model-card	"What is this model?" (unified metadata schema — type, status, risk level, metrics, weights hash)
nanda-model-integrity-layer	"Does this model's metadata meet policy?" (rule-based checks)
nanda-model-governance	"Has this model been cryptographically approved for deployment?" (approval flow with signatures, quorum, scoping, revocation)
nanda-bridge	"How do I expose this to the NANDA network?" (FastAPI router, AgentFacts models, delta sync)

Development

git clone https://github.com/Sharathvc23/nanda-model-provenance.git
cd nanda-model-provenance
pip install -e ".[dev]"
pytest tests/ -v

Credits

Developed by stellarminds.ai,

License

MIT