# LangChain 1.4.0a3

- Product: LangChain (https://whatsnew.fyi/product/langchain)
- Vendor: LangChain
- Date: 2026-09-01
- Version: 1.4.0a3
- Original notes: https://github.com/langchain-ai/langchain/releases/tag/langchain%3D%3D1.4.0a3
- Permalink: https://whatsnew.fyi/product/langchain/releases/1.4.0a3

What's New is an index, not a publisher: every entry below links to the vendor's own release notes, which are the authoritative source. Entries are labelled where they are hand-curated sample data, pre-releases, or drawn from a secondary source such as a developer blog.

Reuse: the summaries, labels and curation here are © What's New. Quote freely with attribution and a link back; wholesale republication of the corpus is not permitted — terms: https://whatsnew.fyi/terms. The vendors' own release notes remain their publishers'.

---

- **added** — Introduce langchain.mcp namespace for adapting MCP servers into LangChain tools
- **added** — Add MCPAdapter class to adapt fastmcp.Client targets including URLs, local scripts, in-process servers, MCPConfig, pre-built clients, and FastMCP ClientGroup
- **added** — Add MCPAdapter.list_tools() method with cache_mode parameter supporting 'use', 'refresh', and 'bypass' options for client-side response caching
- **added** — Add as_langchain_tool() function to convert individual MCP tools for callers managing their own client
- **added** — Add tool metadata grouping under mcp namespace including tool annotations, _meta, and serving server identity
- **added** — Add elicitation='interrupt' parameter to surface server mid-call questions as LangGraph interrupts for human response and run resumption

Third alpha of the `1.4.0` line. This release focuses on the new `langchain.mcp` namespace for adapting MCP servers into LangChain tools.

##### `langchain.mcp` highlights

- **`MCPAdapter`** adapts any target `fastmcp.Client` accepts — a URL, a local script, an in-process server, an `MCPConfig` naming several servers, or a pre-built client — as well as a FastMCP `ClientGroup` for a fleet of servers behind one client.
- **`MCPAdapter.list_tools(*, cache_mode="use")`** discovers and adapts tools, with optional client-side response caching (SEP-2549): `use` serves a cached list within the server's TTL hint, `refresh` repopulates it, `bypass` skips it.
- **`as_langchain_tool(tool, client, *, elicitation=None)`** converts a single MCP tool for callers managing their own client.
- **Tool metadata** is grouped under an `mcp` namespace on each tool: the tool's `annotations` (snake_case) and `_meta` under `metadata["mcp"]["tool"]`, and the serving server's identity under `metadata["mcp"]["server"]`.
- **`elicitation="interrupt"`** surfaces a server's mid-call questions as LangGraph interrupts, so a human answers and the run resumes.

```python
from langchain.agents import create_agent
from langchain.mcp import MCPAdapter

async with MCPAdapter("https://example.com/mcp") as adapter:
    agent = create_agent("anthropic:claude-sonnet-5", await adapter.list_tools())
```

Requires the `mcp` extra: `pip install "langchain[mcp]"` (and `fastmcp>=4.0.0`).

##### Install

```bash
pip install --pre "langchain==1.4.0a3"
```

This is a pre-release; install with `--pre`.

---

*Release notes curated with the assistance of an AI agent.*