OwnCall

MCP-powered Slack bot for interactive Grafana/Prometheus alert investigation.

OwnCall connects to Grafana MCP (and optionally GitHub MCP) and gives your team an on-call assistant directly in Slack.

Features

• @mention investigation — mention the bot in any channel or thread to ask questions about your Grafana dashboards, Loki logs, or Prometheus metrics interactively.
• Alert auto-investigation — when Grafana Alerting or Alertmanager posts a firing alert to Slack, the bot automatically investigates it and replies in the thread.
• YAML-driven prompts — customise the system prompt and add constraints (e.g. "only search the last 3 hours") in config.yml without touching code.
• Multi-MCP support — Grafana MCP is built-in; GitHub MCP can be enabled via a single environment variable to include source code context in investigations.
• Socket Mode — no public endpoint required; runs entirely behind your firewall.

Requirements

• Python 3.10+
• OpenAI API key
• Grafana MCP server (Docker image: grafana/mcp-grafana)
• Slack app with Socket Mode enabled (see Slack App Setup)

Quick Start

1. Install

pip install -e .

2. Configure

Copy the example config and fill in your credentials:

cp config.example.yml config.yml

Edit config.yml:

slack:
  bot_token: "xoxb-..."
  app_token: "xapp-..."

llm:
  model: "gpt-5.4-mini"

agent:
  system_prompt: |
    You are an SRE assistant. Use MCP tools to investigate Grafana alerts.
  constraints:
    - "Search only the last 3 hours unless a time range is specified."

mcp_servers:
  - name: "grafana"
    type: "sse"
    url: "http://localhost:8000/sse"
    enabled: true

3. Start the Grafana MCP server

docker run -p 8000:8000 \
  -e GRAFANA_URL=https://your-grafana.example.com \
  -e GRAFANA_SERVICE_ACCOUNT_TOKEN=your-token \
  grafana/mcp-grafana -t sse

4. Start the bot

export OPENAI_API_KEY=sk-...
owncall -c config.yml

Docker Compose (all-in-one)

cp config.example.yml config.yml  # fill in agent and alert_detection sections
export OPENAI_API_KEY=sk-...
export SLACK_BOT_TOKEN=xoxb-...
export SLACK_APP_TOKEN=xapp-...
export GRAFANA_URL=https://your-grafana.example.com
export GRAFANA_SERVICE_ACCOUNT_TOKEN=your-token
docker compose up

Slack App Setup

1. Go to api.slack.com/apps → Create New App → From scratch.
2. Socket Mode tab → Enable Socket Mode → generate an App-Level Token with connections:write scope (xapp-...).
3. OAuth & Permissions → add Bot Token Scopes:
   • app_mentions:read
   • channels:history
   • chat:write
   • files:write
   • reactions:write
   • groups:history (for private channels)
4. Event Subscriptions → Enable Events → Subscribe to bot events:
   • app_mention
   • message.channels
   • message.groups
5. Install the app to your workspace and copy the Bot User OAuth Token (xoxb-...).

GitHub MCP (optional)

To enable source code lookup during investigations, run a GitHub MCP server and set the following in your environment or config.yml:

export GITHUB_MCP_ENABLED=true
export GITHUB_MCP_URL=http://localhost:9000/sse

Then start a GitHub MCP server (see the docker-compose.yml for an example).

Configuration Reference

Key	Default	Description
slack.bot_token	—	Slack Bot Token (xoxb-...)
slack.app_token	—	Slack App-Level Token (xapp-...)
llm.model	gpt-5.4-mini	OpenAI model name
llm.temperature	0.2	Sampling temperature
agent.system_prompt	—	Base system prompt for the agent
agent.constraints	[]	Additional rules appended to the system prompt
mcp_servers[].name	—	Human-readable server name
mcp_servers[].type	sse	Transport: sse, streamable_http, stdio
mcp_servers[].url	—	SSE endpoint URL
mcp_servers[].enabled	true	Set to false or ${ENV_VAR:-false} to disable
alert_detection.enabled	true	Toggle alert auto-investigation
alert_detection.channels	[]	Restrict to specific channel IDs (empty = all)
alert_detection.rules	—	List of detection rules (see config.example.yml)
mention.channels	[]	Restrict mention responses to specific channel IDs (empty = all)
response.max_length	3000	Max characters per message; longer responses are uploaded as files
response.reaction_on_start	eyes	Reaction added while processing
response.reaction_on_complete	white_check_mark	Reaction added on completion

License

MIT