Mattermost Agents

Mattermost Agents adds AI assistants to a Mattermost workspace for teams working through long discussions and meeting recordings. Teams can use models on their own infrastructure or connect to a cloud provider.

• Discussion summaries: Summarize threads and channels and extract action items from conversations.
• Meeting transcription: Turn meeting recordings into transcripts and summaries.
• Natural language search: Find relevant content across your Mattermost instance by describing what you're looking for.
• Specialized assistants: Create agents with different roles and capabilities, chat with them directly or tag them in an existing conversation.
• Workflow tools: Connect agents to Playbooks and internal systems through MCP, with user approval before tools run.

The plugin is open source under Apache 2.0 and runs on a self-hosted Mattermost server. It supports Ollama, vLLM and OpenAI-compatible APIs, as well as OpenAI, Anthropic, Azure and Amazon Bedrock. Requests go to the chosen model provider: a local backend can keep collaboration data within your network, while cloud providers receive the requests sent to them. The plugin needs network access to its provider.

Access controls govern who can use agents, and token accounting tracks model usage. Semantic search requires PostgreSQL with the pgvector extension.