Answer infrastructure questions in Mattermost with OpenRouter and Qdrant

Overview

AI-powered sub-workflow that answers questions about a your infrastructure configuration directly in a Mattermost channel or thread

Requirements

• OpenRouter/OpenAI/Anthropic API key
• Google Gemini API key — for embeddings
• Jira API credentials — Cloud or Server.
• Mattermost API credentials — to post the reply back to the channel
• Qdrant instance
• Remote MCP servers (see MCP section)
• A sub-workflow that analyses attachments
• A parent workflow that triggers this one via "Execute Workflow" with a properly shaped payload

How it works

1. The workflow is triggered by another workflow
2. ReadIncidentContext logs the request and forwards the payload
3. Call 'attachmentsAnalyzer invokes a vision sub-workflow with the file_ids
4. SetVars sets workflow-level constants
5. AI agent generates a response based on the system prompt, knowledge base, and access to repositories.
6. The agent's response is posted back to Mattermost via the Post a message node

How to use

• Upload your infrastructure documentation (Markdown, YAML, runbooks) into Qdrant
• Import the attachmentsAnalyzer sub-workflow and update the workflow
  reference inside
• Deploy or point to your MCP servers
• Configure credentials
• Edit the AI Agent system message to describe your infrastructure
• In SetVars, replace the example
• Wire this workflow up to an upstream router