silvanexum

Official Python SDK for Silvanexum — the spend-control plane for AI agents. Meter, attribute, cap, and audit what every agent spends across your whole agent-to-agent mesh, on any model or rail. Models are generated from the API's Zod contracts (pydantic v2), so they never drift from the live API.

📚 Full docs: https://docs.silvanexum.com

Install

pip install silvanexum
# uv add silvanexum · poetry add silvanexum

Requires Python ≥ 3.9.

Quickstart — cap spend, then see who spent it

import os
from silvanexum import Silvanexum

sx = Silvanexum(api_key=os.environ["SILVANEXUM_API_KEY"])

# Cap runaway spend — a run or mesh hop that would breach the budget is
# rejected server-side (HTTP 402). "Spent" is computed from real run telemetry.
sx.budgets.create(
    name="support-team",
    scope_type="agent",
    scope_id="agt_123",
    period_type="monthly",
    limit_credits=50_000,
    enforcement="block",
)

# Attribute cost across the mesh — which agent / project / model actually spent.
for row in sx.spending.summary(group_by="agent"):
    print(row.label, row.costCredits)

The client is a context manager, so connections are cleaned up for you:

with Silvanexum() as sx:          # reads SILVANEXUM_API_KEY from the env
    trend = sx.spending.trends(bucket="1d")  # daily spend, last 30 days

What's in the SDK

…plus marketplace, wallet, routing, connectors, team, and more. Full reference: https://docs.silvanexum.com/docs/sdk-reference/python.

Zero-code capture (gateway)

Point your OpenAI client's base_url at the gateway and use your Silvanexum key. Every call is metered on your own provider key and counts toward budgets:

from openai import OpenAI

openai = OpenAI(
    base_url=sx.gateway.openai_base_url(),
    api_key=os.environ["SILVANEXUM_API_KEY"],
    default_headers={"x-silvanexum-trace": trace_id},  # optional mesh stitching
)

Errors

from silvanexum import InsufficientCreditsError, RateLimitError

try:
    sx.runs.create(agent_id="agt_123", prompt="…")
except InsufficientCreditsError:
    ...  # budget/credits exhausted (HTTP 402) — top up or raise the cap
except RateLimitError:
    ...  # client already retried with backoff

Licensed Apache-2.0.