Python: [Feature]: Official first-class TypeScript SDK support
メンテナーはふだん 1 日以内に返信
@rogerbarreto がすでに取り組んでいます。
2026年10月8日 から。
評価
この issue はまだ評価されていません。
説明
Problem
Microsoft Agent Framework currently provides official SDKs for .NET and Python, but not TypeScript. For applications whose frontend, backend, shared domain models, tooling, and deployment are primarily TypeScript, adopting Agent Framework requires introducing and operating a separate .NET or Python service.
That adds another runtime, dependency ecosystem, build pipeline, deployment unit, security-patching process, and set of duplicated contracts. It also makes it harder to share validated types and business logic between the existing application and its agents.
This is related to #1216, but this request is specifically for an official, first-class TypeScript implementation: authored and maintained as TypeScript, distributed through npm with native typed APIs, and treated as a supported Agent Framework SDK rather than only a generic JavaScript runtime package or protocol client.
Use Case and Evidence
We are evaluating Agent Framework for an application that is predominantly TypeScript. The application already uses TypeScript across its UI and server components, and its developers, domain models, validation schemas, tests, and operational tooling are centered on the Node.js ecosystem.
To use Agent Framework today, we would need to:
- Build a separate Python or .NET agent service.
- Recreate or generate contracts between TypeScript and the agent service.
- Operate an additional runtime and deployment lifecycle.
- Duplicate authentication, authorization, telemetry, error handling, and configuration integration.
- Split debugging and local development across language boundaries.
These costs can determine whether a team adopts Agent Framework or chooses a TypeScript-native alternative, even when Agent Framework otherwise provides the preferred capabilities.
Desired Outcome
Provide an officially supported TypeScript SDK that allows Node.js and TypeScript applications to build Agent Framework agents and workflows without introducing a Python or .NET service.
A successful outcome would provide idiomatic, strongly typed TypeScript APIs for the core framework capabilities, with clear compatibility and feature-parity documentation relative to the .NET and Python SDKs. A phased implementation would be valuable even if complete parity cannot be delivered initially.
Alternatives Considered
- Python or .NET sidecar service: Technically workable, but adds operational complexity and prevents direct reuse of TypeScript domain code and types.
- AG-UI, A2A, or custom HTTP integration: Useful for connecting a TypeScript client to a remote agent, but does not let TypeScript developers author Agent Framework agents and workflows natively.
- Direct model-provider SDKs: Support model calls and tool calling, but do not provide the same Agent Framework abstractions and cross-language consistency.
- Other TypeScript-native agent frameworks: Avoid the additional runtime but mean giving up Agent Framework compatibility and Microsoft's unified framework direction.
Proposed Direction
Consider an official npm package, such as @microsoft/agent-framework, initially covering a focused set of core capabilities:
- Agents, messages, tools, structured output, and streaming
- Workflow composition and agent handoffs
- Session and conversation-state abstractions
- Middleware and observability hooks
- MCP, A2A, and AG-UI interoperability
- Microsoft Foundry and common model-provider integrations
Publishing a support matrix and roadmap would help teams adopt an initial release while understanding differences from the .NET and Python implementations.
Scope and Non-Goals
- This request does not require immediate one-to-one parity with every .NET and Python integration.
- A modern Node.js/TypeScript baseline and phased delivery would be reasonable.
- This is not only a request for generated type declarations over an unrelated JavaScript client; the goal is an officially maintained TypeScript Agent Framework SDK with idiomatic APIs and a documented compatibility commitment.
Language
Other / Not Applicable
Contribution Intent
I am willing to implement this
Acknowledgements
- I searched existing issues and did not find a duplicate.
- I will wait for explicit maintainer agreement before starting implementation of a non-trivial change.
- 主要言語
- Python
- スター
- 14k
- フォーク
- 2.5k
- 平均マージ
- 2日 9時間
- マージ済み PR(30日)
- 434
環境構築
このプロジェクトの開発コンテナを、あなたの GitHub アカウントでブラウザ上に起動します。
- Dockerfile・Docker Compose ファイルなし
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
microsoft/agent-framework のほかの issue
-
Python: [Bug]: A non-positive MCP request timeout silently disables the timeout対応中かも @jpalvarezl が 4 日前に担当しました。 オープンpython
難易度 2/5 1〜3時間 初心者へのやさしさ 82/100
microsoft/agent-framework#9313 · 担当者 1 名 ·
メンテナーはふだん 1 日以内に返信
-
.NET python triage
難易度 1/5 1時間未満 初心者へのやさしさ 85/100
microsoft/agent-framework#9092 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
-
Python: raw-data content mappings lose annotations and attachment metadata対応中かも @moonbox3 が 18 日前に担当しました。 オープンpython triage
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
microsoft/agent-framework#8632 · コメント 2 件 ·
メンテナーはふだん 1 日以内に返信
-
Python: Clarify when to use platform対応中かも @eavanvalkenburg が 17 日前に担当しました。 オープンpython triage
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
microsoft/agent-framework#8599 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
-
.NET compaction documentation
難易度 1/5 1時間未満 初心者へのやさしさ 82/100
microsoft/agent-framework#4629 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
microsoft/agent-framework の issue をすべて見る
似ている issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 82/100
sierra-research/tau2-bench#635 ·
メンテナーはふだん 1 日以内に返信
-
The add-ssm-model skill requires a Cython simulator; the contributing guide allows Python ones対応中かも @fmuia が今日担当しました。 オープンdocumentation skill
難易度 2/5 1〜3時間 初心者へのやさしさ 83/100
lnccbrown/ssm-simulators#406 ·
メンテナーはふだん 4 日以内に返信
-
難易度 1/5 1時間未満 初心者へのやさしさ 95/100
btclib-org/btclib-ecc#309 ·
メンテナーはふだん 1 日以内に返信
-
bug
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
jbacalso24/mimry#459 · コメント 1 件 ·
メンテナーはふだん 1 日以内に返信
-
[Bug] POST /revoke with a FastMCP access token is a silent no-op (ownership gate never matches)オープンauth bug high-priority
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
メンテナーはふだん 1 日以内に返信