Investigate Mattermost incident reports with an AI SRE assistant using OpenRouter

Overview

AI-powered SRE sub-workflow that investigates user-reported incidents coming from a Mattermost channel and posts a structured diagnostic report back into the same thread. The result is a four-section incident report:

• What happened — a concise summary of the symptoms
• Event timeline — what was happening in the 10+ minutes before the incident
• Root cause — up to two probable causes
• Troubleshooting tips — step-by-step remediation for each root cause

Requirements

• OpenRouter/OpenAI/Anthropic API key
• Google Gemini API key — for embeddings
• 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 (When Executed by Another Workflow)
2. ReadIncidentContext logs the incoming classification for debugging and forwards the payload downstream
3. Call 'attachmentsAnalyzer invokes a vision sub-workflow with the file_ids
4. SetVars defines the configuration used by the AI Agent prompt
5. AI Agent runs the investigation
6. Post a message` sends the agent's final output back to the originating channel

How to use

• Prepare the knowledge base — upload your infrastructure description, service map, naming conventions and runbooks into a Qdrant collection
• Deploy the MCP servers and put their URLs into the corresponding tool
• Configure credentials
• Adjust the configuration in SetVars
• Plug in the attachments sub-workflow — replace the workflow reference in Call 'attachmentsAnalyzer' with your own vision-analysis workflow that maps file_ids[]
• Tune the system prompt in the AI Agent — add project-specific naming conventions, ownership info, escalation rules and any quirks of your infrastructure

• Connect classification workflow as parent