teams-anthropic-integration_skill

This skill helps you integrate Anthropic Claude models into Teams.ai apps, enabling web search with MCP when needed for up-to-date results.

TypeScript

4

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill youdotcom-oss/agent-skills --skill teams-anthropic-integration

SKILL.md11.3 KB

Overview

This skill adds Anthropic Claude models (Opus, Sonnet, Haiku) to Microsoft Teams.ai applications using @youdotcom-oss/teams-anthropic. It provides a clear Basic Setup for Claude-powered Teams bots and an optional MCP integration path to add You.com web search and content extraction. The goal is fast integration, streaming responses, function calling, and optional real-time web access.

How this skill works

Install the teams-anthropic package, configure your Anthropic API key, and create an AnthropicChatModel to handle conversations inside Teams.ai. For web search and page extraction, add the McpClientPlugin and point it at the You.com MCP endpoint with a YDC_API_KEY; use ChatPrompt to combine Claude responses with MCP results. The skill supports streaming tokens, function calling, and local conversation memory for richer bot interactions.

When to use it

Building a Microsoft Teams bot that uses Anthropic Claude models.
Adding Claude (Opus, Sonnet, Haiku) to an existing Teams.ai app.
Needing streaming responses or function calling from Claude inside Teams.
Wanting fresh web search results and content extraction via You.com MCP.
Prototyping or productionizing an enterprise Teams AI assistant.

Best practices

Start with Path A (Basic Setup) to validate Claude integration before adding MCP.
Use AnthropicModel.CLAUDE_SONNET_4_5 for balanced performance; choose Opus for complex tasks and Haiku for low-latency use cases.
Keep API keys in environment variables (ANTHROPIC_API_KEY, YDC_API_KEY) and validate at startup.
Use ChatPrompt + McpClientPlugin only when web search is required to avoid extra dependencies and network complexity.
Stream tokens to the Teams client with onChunk for better UX on long responses.

Example use cases

Customer support bot in Teams that answers product questions and calls backend functions for orders.
Internal knowledge assistant that searches the web and extracts contract clauses via MCP.
Developer helper in Teams that pulls live documentation and code examples for queries like React APIs.
Meeting assistant that streams summaries during calls and remembers context with LocalMemory.
Operational bot that triggers function calls (e.g., get_weather) and returns structured results in chat.

FAQ

Use Sonnet (AnthropicModel.CLAUDE_SONNET_4_5) for balanced speed and capability. Choose Opus for complex reasoning and Haiku for efficiency and low latency.

Do I need the You.com MCP integration?

No. Path A works for chat, streaming, and function calling. Add MCP only if you need web search or content extraction.

How do I handle missing modules or keys?

Install required packages (teams-anthropic, anthropic SDK, teams.mcpclient) and set ANTHROPIC_API_KEY and YDC_API_KEY in .env. Check install errors and logger output for details.