Skip to main content

parmana-connector-sdk

Contracts and reference implementations for building a Connector for Parmana, the authorization layer for AI execution.

A Connector is the last hop in Parmana's pipeline: it receives an already-authorized, already-verified request and carries it out against a real system (an API, a database, anything). A Connector never evaluates policy, never authorizes execution, and never resolves its own credentials — it only executes, using a credential Parmana has already resolved for it.

This is the Python counterpart to the TypeScript @parmana/connector-sdk package — same contract, same design, idiomatic Python (synchronous, dataclasses and Protocol instead of interfaces).

Install

pip install parmana-connector-sdk

The contract

Every Connector implements one protocol:

from parmana_connector_sdk import (
    Connector,
    ConnectorRequest,
    ConnectorExecutionContext,
    ConnectorResponse,
    connector_capabilities,
)


class MyConnector:
    connector_id = "my-service"
    capabilities = connector_capabilities(["my-service:do-thing"])

    def execute(
        self, request: ConnectorRequest, context: ConnectorExecutionContext
    ) -> ConnectorResponse:
        # request.parameters is exactly what was authorized -- never re-derive it.
        # context.credential is an opaque, already-resolved CredentialHandle.
        # context.timeout_ms is how long you have.
        return ConnectorResponse(success=True, metadata={})

Connector is a typing.Protocol — MyConnector doesn't need to subclass anything, it just needs to match the shape (connector_id, capabilities, execute).

capabilities must be namespaced verbs ("namespace:verb", e.g. "crm:update-contact"), validated eagerly by connector_capabilities() so a malformed capability fails at construction, not mid-execution.

One reference implementation, for testing

MockConnector gives you scripted responses or failure injection, and records every request it receives, so you can test your own Connector's callers without standing up a real endpoint:

from parmana_connector_sdk import MockConnector, MockConnectorOptions, MockConnectorScript, connector_capabilities

connector = MockConnector(
    MockConnectorOptions(
        connector_id="billing",
        capabilities=connector_capabilities(["billing:issue-refund"]),
        script=MockConnectorScript(
            respond=lambda request, context: ConnectorResponse(
                success=True, metadata={"refund_id": "r_1"}
            )
        ),
    )
)

connector.execute(request, context)
connector.invocations  # every request it received, for assertions

This package intentionally ships no HTTP-calling reference implementation — only the contract and the hermetic mock above. A REST-shaped connector is usually this simple to write directly against the contract:

import urllib.request
import json

from parmana_connector_sdk import (
    ConnectorRequest,
    ConnectorExecutionContext,
    ConnectorResponse,
    connector_capabilities,
)


class BillingConnector:
    connector_id = "billing"
    capabilities = connector_capabilities(["billing:issue-refund"])

    def __init__(self, base_url: str) -> None:
        self.base_url = base_url

    def execute(
        self, request: ConnectorRequest, context: ConnectorExecutionContext
    ) -> ConnectorResponse:
        token = context.credential.value  # shape is whatever your CredentialProvider resolved
        body = json.dumps(request.parameters).encode()
        req = urllib.request.Request(
            f"{self.base_url}/{request.action}",
            data=body,
            headers={"Authorization": f"Bearer {token}", "Content-Type": "application/json"},
            method="POST",
        )
        with urllib.request.urlopen(req, timeout=context.timeout_ms / 1000) as response:
            if response.status >= 300:
                raise RuntimeError(f"HTTP {response.status}")
            return ConnectorResponse(success=True, metadata=json.load(response))

Four worked examples

parmana_connector_sdk.connectors.{oracle,salesforce,sap,workday} are deterministic MockConnectors shaped like a real enterprise integration — each declares one realistic capability and nothing else. They are explicit mocks, not real integrations; read them as a template for structuring your own connector's capability and metadata, not as something to call in production.

Credentials

A Connector never resolves its own credentials. CredentialProvider is the seam Parmana's Execution Gateway uses upstream of a Connector — EnvironmentCredentialProvider (reads a mapped env var) and StaticCredentialProvider (in-memory, for tests) are the two provided implementations.

Full reference

See the Connector SDK reference and the Connector Development Guide — written against the TypeScript package, but the contract this package implements is identical.

Running the tests

The tests import the package directly from this folder, so put it on the import path first. Run these from the repository root.

Windows (PowerShell):

$env:PYTHONPATH = "python-connector-sdk"
python -m pytest python-connector-sdk/tests

macOS or Linux:

PYTHONPATH=python-connector-sdk python -m pytest python-connector-sdk/tests

Install pytest first if needed: pip install pytest. Without PYTHONPATH, collection fails with ModuleNotFoundError: No module named 'parmana_connector_sdk'.

Metadata

Release files for parmana-connector-sdk 0.2.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for parmana-connector-sdk 0.2.1
File Size Uploaded Provenance Actions
parmana_connector_sdk-0.2.1.tar.gz 18.6 kB Without provenance Details for parmana_connector_sdk-0.2.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for parmana-connector-sdk 0.2.1
File Interpreter ABI Platform Actions
parmana_connector_sdk-0.2.1-py3-none-any.whl Python 3 none any Details for parmana_connector_sdk-0.2.1-py3-none-any.whl

Total release size: 36.4 kB

Release files / parmana_connector_sdk-0.2.1.tar.gz

Download URL parmana_connector_sdk-0.2.1.tar.gz
Size 18.6 kB
Tags Source
SHA-256 checksum
How to use checksums
458dc4b58bd458b8df349ce3bda4aca21a177c35cbbce8108661cc8c557cb31d
BLAKE2b-256 checksum
How to use checksums
b6601c5f17e1627533020123ed37ca8542a4accd37c3b2e905c2c91269d76abb
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.6

Release files / parmana_connector_sdk-0.2.1-py3-none-any.whl

Download URL parmana_connector_sdk-0.2.1-py3-none-any.whl
Size 17.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
f6f5dfbc02290bd0a8f9adb0bf5525b25a1d8fecc9aae04aa04bfa29fcba3b50
BLAKE2b-256 checksum
How to use checksums
1d6d93f932535901ae7d1702a4501409d7d585be4307870d264bf20b9a914e58
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.6

Security for parmana-connector-sdk 0.2.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page