palveron-sdk

Official Python SDK for the Palveron AI Governance Gateway — policy enforcement, trace verification, and blockchain-anchored audit trails for every AI interaction.

Every AI interaction your application makes — governed, audited, and optionally anchored to the blockchain. In one import.

• One dependency (httpx) — clean install, no surprises
• Sync + Async — Palveron and AsyncPalveron clients with the same surface
• Multi-modal — text, images, audio, documents, code
• Enterprise-grade — retry with exponential backoff, circuit breaker, typed errors
• On-prem ready — point to any Palveron Gateway endpoint
• Typed — full type hints, ships with py.typed

Installation

pip install palveron-sdk

Quick Start

from palveron import Palveron

client = Palveron(api_key="pv_live_xxx")

result = client.verify("Transfer $50,000 to account DE89370400440532013000")

if result.is_blocked:
    raise RuntimeError(f"Blocked by policy: {result.reason}")

# result.output is the (possibly sanitized) prompt — always use it
# instead of the raw input so downstream LLMs never see PII / secrets.
print(result.output, result.trace_id)

Async Support

import asyncio
from palveron import AsyncPalveron

async def main():
    async with AsyncPalveron(api_key="pv_live_xxx") as client:
        result = await client.verify("Is this safe?")
        print(result.decision)

asyncio.run(main())

Features

• Policy Enforcement — every prompt routed through your active guardrails before it reaches an LLM
• Trace Verification — every decision logged with an integrity hash for tamper detection
• Multi-modal attachments — Attachment.from_file() and Attachment.from_bytes() with auto MIME detection
• Agentic / MCP context — pass RequestContext so the audit trail records the tool chain, not just the prompt
• Blockchain Attestation — high-severity traces anchored to Flare for cryptographic audit trails
• EU AI Act / DORA / GDPR / NIST AI RMF — compliance-ready audit fields out of the box

Configuration

client = Palveron(
    api_key="pv_live_xxx",                   # Required — project or agent API key
    base_url="https://gateway.palveron.com", # Custom endpoint for on-prem
    timeout=30.0,                            # Request timeout in seconds
    max_retries=3,                           # Retry attempts on transient failures
    headers={"X-Tenant": "acme"},            # Custom headers on every request
    circuit_threshold=5,                     # Failures before circuit opens
    circuit_cooldown=30.0,                   # Cooldown before half-open retry (seconds)
)

API Reference

Full reference at docs.palveron.com/sdks. Quick summary:

Examples

LLM gateway with governance (OpenAI)

from palveron import Palveron
from openai import OpenAI

palveron = Palveron(api_key=os.environ["PALVERON_API_KEY"])
openai = OpenAI()

def ask_with_governance(user_prompt: str) -> str:
    gate = palveron.verify(user_prompt)
    if gate.is_blocked:
        raise RuntimeError(f"Blocked by policy: {gate.reason}")
    # Always use gate.output instead of the raw input so the LLM never
    # sees PII / secrets the gateway redacted.
    resp = openai.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": gate.output}],
    )
    return resp.choices[0].message.content

Agentic / MCP audit context

from palveron import Palveron, RequestContext

client = Palveron(api_key=os.environ["PALVERON_API_KEY"])

result = client.verify(
    "Execute bank transfer",
    context=RequestContext(
        mcp_server="https://banking-mcp.corp.internal",
        tool_name="transfer_funds",
        chain_depth=3,
        source_system="crewai",
        session_id="agent_session_42",
    ),
)

On-premise / self-hosted gateway

client = Palveron(
    api_key=os.environ["PALVERON_API_KEY"],
    base_url="https://gateway.internal.acme.corp:8080",
    timeout=10.0,
    max_retries=5,
)

Error Handling

All errors extend PalveronError with structured metadata:

from palveron import Palveron, PalveronError, PalveronRateLimitError
import time

client = Palveron(api_key=os.environ["PALVERON_API_KEY"])

try:
    client.verify(user_input)
except PalveronRateLimitError as err:
    # err.retry_after_ms — wait this long before retrying
    time.sleep(err.retry_after_ms / 1000)
except PalveronError as err:
    print(err.code, err.status_code, err.request_id)

Requirements

• Python 3.8 or newer
• Single runtime dependency: httpx

Links

• Documentation — docs.palveron.com
• SDK reference — docs.palveron.com/sdks
• Dashboard — palveron.com
• Support — hello@palveron.com
• GitHub — palveron/sdk-python
• Changelog — CHANGELOG.md

License

MIT — Copyright © 2026 Palveron.