ToolRegistry

English Version | 中文版

A protocol-agnostic tool management library for function-calling LLMs.

Ecosystem

Package	Description	Docs
toolregistry	Core library — tool registration, schema generation, execution	Docs
toolregistry-server	Server adapters — expose tools via OpenAPI & MCP	Docs
toolregistry-hub	Ready-to-use tools — calculator, web search, file ops, etc.	Docs

toolregistry (core)
       ↓
toolregistry-server (tool server)
       ↓
toolregistry-hub (tool collection + server config)

Features

Protocol-agnostic — register tools from native Python functions/classes, MCP servers, OpenAPI specs, or LangChain tools through a unified interface
Multi-provider schemas — generate tool schemas for OpenAI, Anthropic, and Gemini via llm-rosetta
Concurrent execution — thread and process pool backends with per-tool timeout and concurrency control
Permission system — tag-based policies (READ_ONLY, DESTRUCTIVE, NETWORK, etc.) with allow/deny/ask rules
Tool metadata & tags — classify tools with ToolTag, ToolMetadata, namespace support, and source tracking
Admin panel — built-in Web UI for monitoring tools, permissions, and runtime config (i18n: EN/ZH)
Think-augmented calling — inject chain-of-thought reasoning into tool calls (arXiv:2601.18282)
Declarative config — load tool sources from JSONC/YAML config files
Zero-dependency core — HTTP client, YAML parser, JSON Schema resolver all vendored; only pydantic and llm-rosetta as runtime deps

Quick Start

from toolregistry import ToolRegistry

registry = ToolRegistry()

@registry.register
def add(a: float, b: float) -> float:
    """Add two numbers together."""
    return a + b

# Use with any LLM provider
schemas = registry.get_schemas(api_format="openai-chat")  # or "anthropic", "gemini"
result = registry["add"](1, 2)  # 3.0

See the Usage Guide for MCP, OpenAPI, LangChain integrations and more.

Installation

Requires Python >= 3.10.

pip install toolregistry                   # core
pip install toolregistry[mcp]              # + MCP support
pip install toolregistry[langchain]        # + LangChain support
pip install toolregistry-hub               # ready-to-use tools (separate package)

Citation

@software{toolregistry2025,
  title={ToolRegistry: A Protocol-Agnostic Tool Management Library for OpenAI-Compatible LLM Applications},
  author={Peng Ding},
  year={2025},
  url={https://github.com/Oaklight/ToolRegistry},
  note={A Python library for unified tool registration, execution, and management across multiple protocols in OpenAI-compatible LLM applications}
}

@article{ding2025toolregistry,
  title={Toolregistry: A protocol-agnostic tool management library for function-calling llms},
  author={Ding, Peng and Stevens, Rick},
  journal={arXiv preprint arXiv:2507.10593},
  year={2025}
}

License

MIT — see LICENSE for details.

Release files for toolregistry 0.17.0

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

Source distribution (sdist)

Source distribution for toolregistry 0.17.0
File	Size	Uploaded
toolregistry-0.17.0.tar.gz	356.4 kB	Sep 10, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for toolregistry 0.17.0
File	Interpreter	ABI	Platform	Reset
toolregistry-0.17.0-py3-none-any.whl 276.1 kB Sep 10, 2026	Python 3	none	any	Details

Total release size: 632.5 kB

Release files / toolregistry-0.17.0.tar.gz

Download URL	toolregistry-0.17.0.tar.gz
Size	356.4 kB
Tags	Source
SHA-256 checksum How to use checksums	`4ffa2b8e2522572361701deb542c278b6465fa02d12876e7cb1631ed543c2ca6`
BLAKE2b-256 checksum How to use checksums	`27d122d264cbec7112ca6d73fbb9476e410d7f40de87ae42066e01a18b95e351`
Upload date	Sep 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

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.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/Oaklight/ToolRegistry
Publishing commit: github.com/Oaklight/ToolRegistry/tree/1ed9ef450ba899560747e795edb47d5346331650

Workflow

Publishing configuration: github.com/Oaklight/ToolRegistry/blob/1ed9ef450ba899560747e795edb47d5346331650/.github/workflows/release.yml
Publishing logs: github.com/Oaklight/ToolRegistry/actions/runs/34505319664/attempts/1

Artifact

Subject: toolregistry-0.17.0.tar.gz
SHA-256 checksum: 4ffa2b8e2522572361701deb542c278b6465fa02d12876e7cb1631ed543c2ca6 Verifying attestations

Release files / toolregistry-0.17.0-py3-none-any.whl

Download URL	toolregistry-0.17.0-py3-none-any.whl
Size	276.1 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`c416402b34cca6c1a6a9ebfb9e56007756ce794d4282c39efab6c3857d67f4f5`
BLAKE2b-256 checksum How to use checksums	`6932763b1d48f65e52760225506ecf0edb049f2558e7f4daeb485ad5e9f6d4a7`
Upload date	Sep 10, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/Oaklight/ToolRegistry
Publishing commit: github.com/Oaklight/ToolRegistry/tree/1ed9ef450ba899560747e795edb47d5346331650

Workflow

Publishing configuration: github.com/Oaklight/ToolRegistry/blob/1ed9ef450ba899560747e795edb47d5346331650/.github/workflows/release.yml
Publishing logs: github.com/Oaklight/ToolRegistry/actions/runs/34505319664/attempts/1

Artifact

Subject: toolregistry-0.17.0-py3-none-any.whl
SHA-256 checksum: c416402b34cca6c1a6a9ebfb9e56007756ce794d4282c39efab6c3857d67f4f5 Verifying attestations