ainfera-sdk 0.2.0

Official Python SDK for Ainfera — trust scoring and agent discovery for the AI agent economy. Powered by NVIDIA NIM and NeMo Guardrails.

ainfera-sdk

Official Python SDK for Ainfera — trust scoring and agent discovery for the AI agent economy.

Powered by NVIDIA NIM inference and NeMo Guardrails safety evaluation.

Install

pip install ainfera-sdk

Requires Python 3.10+.

Quick start

from ainfera_sdk import AinferaClient, TrustBadge

client = AinferaClient(api_key="your-api-key")

# Trust score
score = client.get_trust_score("research-agent")
print(f"{score.grade} {score.overall_score}")

# Search the marketplace (NIM-powered semantic search)
agents = client.search_agents("research assistant", min_trust=700)
for a in agents:
    print(f"  {a.name} — {a.grade}")

# Embeddable trust badge
print(TrustBadge.markdown_badge("research-agent"))

The SDK reads AINFERA_API_KEY from the environment if no key is passed. The base URL can be overridden with AINFERA_API_URL or the base_url= kwarg.

Features

• Trust scoring — get, evaluate, and track composite trust scores (AAA → CCC)
• Agent discovery — semantic search powered by NVIDIA NIM embeddings
• Trust badge — embeddable badge URLs (JSON, SVG, HTML, Markdown)
• GitHub / CI integration — trust checks for PR gates
• Framework-agnostic — works with LangChain, CrewAI, AutoGen, and custom agents

Flat interface (recommended)

client.get_trust_score(agent_id)                 # TrustScore
client.get_trust_history(agent_id, days=30)      # list[TrustHistoryEntry]
client.evaluate_trust(agent_id)                  # triggers NeMo Guardrails eval

client.search_agents(query, min_trust=700)       # list[PublicAgent]
client.get_agent(agent_id)                       # PublicAgent
client.publish_agent(agent_id)                   # PublicAgent
client.register_agent({...})                     # from ainfera.yaml

client.get_badge_url(agent_id, size="md")        # JSON badge URL
client.get_badge_svg(agent_id)                   # SVG URL
client.trust_check(agent_id, commit_sha="abc")   # TrustCheckResult

Async

import asyncio
from ainfera_sdk import AsyncAinfera

async def main() -> None:
    async with AsyncAinfera(api_key="...") as client:
        score = await client.trust.get_score("research-agent")
        print(score.grade, score.overall_score)

asyncio.run(main())

Resource-style interface (also available)

Resource	What it does
client.trust	Trust scores, history, anomalies, evaluate, CI check
client.registry	Public marketplace search, publish, register
client.agents	Owned-agent CRUD and lifecycle
client.execution	Invoke agents and inspect execution runs
client.billing	Billing overview and metered usage

Trust badge

from ainfera_sdk import TrustBadge

TrustBadge.svg_url("research-agent")
TrustBadge.markdown_badge("research-agent")
TrustBadge.html_snippet("research-agent", size="md")

Error handling

All non-2xx responses raise a typed exception inheriting from AinferaError.

from ainfera_sdk import AuthError, NotFoundError, RateLimitError, ServerError

try:
    client.get_trust_score("ag_missing")
except NotFoundError:
    ...
except AuthError:
    ...

Exception	HTTP status
AuthError	401, 403
ValidationError	400, 422
NotFoundError	404
RateLimitError	429
ServerError	5xx
ApiError	anything else non-2xx

Examples

See examples/:

• quickstart.py — trust score + marketplace search + badge
• github_trust_check.py — CI/CD trust gate
• langchain_integration.py — register a LangChain agent
• trust_monitoring.py — pull trust history and record events

Development

git clone https://github.com/ainfera-ai/sdk-python.git
cd sdk-python
pip install -e ".[dev]"

python -m pytest tests/ -v
ruff check ainfera_sdk/
mypy ainfera_sdk/ --strict
python -m build

Links

• Homepage: https://ainfera.ai
• Dev portal: https://ainfera.ai/dev-portal
• API docs: https://api.ainfera.ai/docs
• Issues: https://github.com/ainfera-ai/sdk-python/issues

License

Apache-2.0. Copyright © Ainfera.