Skip to main content
Pre-release

This release is a pre-release and may not be stable for production use.

Swarmauri Logo

PyPI - Downloads Hits PyPI - Python Version PyPI - License PyPI - swarmauri_core Discord

Swarmauri Core

swarmauri_core provides the interface contracts for the Swarmauri SDK. It is the package component authors use when they need stable abstract interfaces for agents, tools, LLMs, parsers, vector stores, cryptography, signing, key providers, transports, middleware, billing, storage, XMP, and other composable intelligence infrastructure components.

Why Swarmauri Core?

swarmauri_core keeps component contracts separate from implementations. Package authors can depend on stable interfaces without pulling in every base class, provider adapter, model package, or runtime integration.

FAQ

Q: What belongs in swarmauri_core?

A: Interface contracts, protocol shapes, enums, and shared types for Swarmauri component families.

Q: Should application developers instantiate classes from this package?

A: Usually no. Most application code uses concrete implementations from swarmauri_standard, community packages, or provider-specific packages. Component authors use swarmauri_core to implement the correct contract.

Q: How does this package relate to swarmauri_base?

A: swarmauri_core defines interfaces. swarmauri_base implements reusable base classes and serialization behavior on top of those interfaces.

What Is Included?

The package is intentionally implementation-light. It defines contracts, protocol shapes, enums, and shared types so implementation packages can interoperate without depending on each other directly.

Major interface families include:

  • Agents and agent APIs: IAgent, conversation, parser, retrieval, toolkit, and vector-store agent mixins.
  • LLM and multimodal prediction: text LLMs, tool LLMs, OCR, STT, TTS, and VLM interfaces.
  • Documents and parsing: document contracts, parser contracts, chunkers, conversations, prompts, and prompt templates.
  • Retrieval and vectors: vector, matrix, tensor, metric, similarity, vector-store comparator, deprecated distance compatibility contracts, vector store, and document store contracts.
  • Security and trust: crypto, MRE crypto, cipher suites, signing, proof-of-possession, token services, key providers, certificates, and XMP embedding.
  • Runtime infrastructure: middleware, transports, storage adapters, publishers, rate limits, tracing, state, service registries, and pipelines.
  • Business systems: billing provider interfaces for customers, payments, invoices, subscriptions, refunds, payouts, risk, reporting, webhooks, and marketplace flows.

Features

  • Stable abstract interfaces for Swarmauri component packages.
  • Shared types for crypto, signing, key management, transports, billing, and proof-of-possession workflows.
  • Zero runtime dependencies.
  • Interface-first design for packages that later inherit reusable behavior from swarmauri_base.
  • Python 3.10, 3.11, 3.12, 3.13, and 3.14 support.

Installation

Install with uv:

uv add swarmauri_core

Install with pip:

pip install swarmauri_core

Usage

Use swarmauri_core when defining a new implementation package or validating that a class satisfies a public Swarmauri contract.

import asyncio
from typing import Any, Optional

from swarmauri_core.agents.IAgent import IAgent
from swarmauri_core.messages.IMessage import IMessage


class MinimalAgent(IAgent):
    def exec(
        self,
        input_data: Optional[Any] = None,
        llm_kwargs: Optional[dict] = None,
    ) -> Any:
        return {"input": input_data, "kwargs": llm_kwargs or {}}

    async def aexec(
        self,
        input_str: Optional[str | IMessage] = "",
        llm_kwargs: Optional[dict] = None,
    ) -> Any:
        return self.exec(input_str, llm_kwargs)

    def batch(
        self,
        inputs: list[str | IMessage],
        llm_kwargs: Optional[dict] = None,
    ) -> list[Any]:
        return [self.exec(item, llm_kwargs) for item in inputs]

    async def abatch(
        self,
        inputs: list[str | IMessage],
        llm_kwargs: Optional[dict] = None,
    ) -> list[Any]:
        return await asyncio.gather(
            *(self.aexec(item, llm_kwargs) for item in inputs)
        )

Implement a parser contract:

from pathlib import Path

from swarmauri_core.documents.IDocument import IDocument
from swarmauri_core.parsers.IParser import IParser


class PlainTextParser(IParser):
    def parse(self, data: str | bytes | Path) -> list[IDocument]:
        raise NotImplementedError("Return concrete Document instances here.")

Implement a storage or retrieval contract in a concrete package, then expose the class through swarmauri_base and the public swarmauri namespace when appropriate.

Component Author Workflow

  1. Choose the closest interface from swarmauri_core.
  2. Implement domain behavior in a package-specific class.
  3. Add reusable serialization and registration behavior with swarmauri_base when the implementation should become a Swarmauri component.
  4. Publish or register the implementation package through the appropriate swarmauri.<kind> entry point or namespace mapping.
  5. Document installation and usage in the implementation package README.

Related Packages

Foundational packages:

  • swarmauri provides the namespace importer and plugin discovery layer.
  • swarmauri_base provides reusable base classes built on these interfaces.
  • swarmauri_standard provides first-party implementations for common component kinds.

Packages built around specific core component kinds:

When To Use This Package

Use swarmauri_core when you are designing or implementing a Swarmauri-compatible component and need the interface contract. Use swarmauri_base when you also need reusable base-class behavior, serialization helpers, and component registration. Use swarmauri_standard when you want ready-to-use implementations.

License

Apache-2.0

Contributing

When adding or changing a core interface, keep the contract narrow, update downstream base classes where needed, add focused tests, and follow the Swarmauri SDK contribution guide.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

swarmauri_core-0.11.0.dev2.tar.gz (64.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

swarmauri_core-0.11.0.dev2-py3-none-any.whl (126.7 kB view details)

Uploaded Python 3

File details

Details for the file swarmauri_core-0.11.0.dev2.tar.gz.

File metadata

  • Download URL: swarmauri_core-0.11.0.dev2.tar.gz
  • Upload date:
  • Size: 64.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.26 {"installer":{"name":"uv","version":"0.11.26","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for swarmauri_core-0.11.0.dev2.tar.gz
Algorithm Hash digest
SHA256 5434eba62c461ca458104a0bc9f683bb84dd55d133b6593b7f33586a64ebc416
MD5 bc7f3e15359ed147dda783b012d35c2c
BLAKE2b-256 cb8a722d266f7b5c8330cccd353f2e6dded78ed92e801260feaf51292709ea63

See more details on using hashes here.

File details

Details for the file swarmauri_core-0.11.0.dev2-py3-none-any.whl.

File metadata

  • Download URL: swarmauri_core-0.11.0.dev2-py3-none-any.whl
  • Upload date:
  • Size: 126.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.11.26 {"installer":{"name":"uv","version":"0.11.26","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for swarmauri_core-0.11.0.dev2-py3-none-any.whl
Algorithm Hash digest
SHA256 0c2aea83696cc9b26a852e6a044389fcd499278cde5881ab38f7dfa3c1b976fa
MD5 ecae748ea594370a78968add03e7ba06
BLAKE2b-256 4abd25a6b6ffa0b06036a0bb87f3d4ac20caab96bfe48bae31402810d827c42a

See more details on using hashes here.

Release history Release notifications | RSS feed

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page