Why
You're in the zone writing code. Then:
- You
git pushand realize there's a SEV-1 in progress - You get paged and scramble between Slack, Datadog, and your incident tool
- The retro after an incident takes hours to compile
This plugin brings Rootly's incident lifecycle into Claude Code -- so you can assess deployment risk, investigate incidents, check on-call, and generate retrospectives from the same terminal where you write code.
What You Get
<table> <tr> <td width="50%">Before You Deploy
> /rootly:deploy-checkAnalyzes your git diff against past incidents. Warns you if similar changes caused outages before. Checks if on-call coverage exists.
</td> <td width="50%">When You Get Paged
> /rootly:respond INC-4521Pulls full incident context, finds similar past incidents, suggests proven solutions, and shows who's on-call -- all in one brief.
</td> </tr> <tr> <td width="50%">During Your Shift
> /rootly:oncallSee who's on-call across all schedules, shift metrics, upcoming handoffs, and health risk indicators.
</td> <td width="50%">Service Health Check
> /rootly:statusQuick overview of all services with active incidents grouped by severity and age.
</td> </tr> <tr> <td width="50%">Stakeholder Communication
> /rootly:brief INC-4521Generate executive-friendly incident summaries with impact, timeline, and current status for stakeholder updates.
</td> <td width="50%">Shift Handoffs
> /rootly:handoffCreate structured handoff documents for incident commanders or on-call transitions with context and next steps.
</td> </tr> <tr> <td width="50%">Ask Questions
> /rootly:ask "incidents this week"Natural language queries about your incident data, on-call schedules, and service reliability patterns.
</td> <td width="50%">After the Dust Settles
> /rootly:retro INC-4521Generates structured retrospectives from incident data: timeline, contributing factors, action items, and systemic patterns.
</td> </tr> </table>Installation
You can use this plugin in two ways:
- Marketplace install for a persistent Claude Code installation
- Local
--plugin-dirloading for development and evaluation from source
Marketplace Install
This repository includes .claude-plugin/marketplace.json, so Claude Code can use the repo itself as a marketplace source.
- Add the marketplace:
/plugin marketplace add Rootly-AI-Labs/rootly-claude-plugin- Open the plugin manager:
/plugin- In the Discover tab, select
rootlyand install it to your preferred scope:
- User: available across all your projects
- Project: shared through this repository's
.claude/settings.json - Local: only for you in this repository
- Reload plugins so the install takes effect immediately:
/reload-plugins- Run setup -- Claude will automatically handle OAuth2 login when it connects to the MCP server:
/rootly:setupA browser window will open for you to log in to Rootly and grant access. No API token needed.
Local Source Loading
Step 1: Clone the Plugin
git clone https://github.com/Rootly-AI-Labs/rootly-claude-plugin.git
cd rootly-claude-pluginStep 2: Load It in Claude Code
claude --plugin-dir .Claude Code loads the plugin directly from this directory for the current session. This is the recommended flow for local development and evaluation. For a persistent install, use the marketplace flow above.
Step 3: Verify
/rootly:setupDirect MCP Access
This repository is a Claude Code plugin. If you only want direct Rootly MCP access in Claude Desktop / Cowork, configure the MCP server separately:
{
"mcpServers": {
"rootly": {
"url": "https://mcp.rootly.com/mcp"
}
}
}Claude will handle OAuth2 login automatically -- a browser window opens for you to authenticate with Rootly. No API token needed.
Setup & Configuration
After installation, run the setup command:
> /rootly:setupFirst-time plugin setup with API token validation, service mapping, and quick-start guide.
Authentication
OAuth2 (Recommended)
MCP commands use OAuth2 automatically. When Claude connects to the Rootly MCP server, it handles the OAuth2 flow -- a browser window opens for you to log in and grant access. No configuration needed.
To re-authenticate, disconnect and reconnect the MCP server via /mcp.
API Token (Hook Scripts)
Hook scripts (active-incident warnings on commit/push) still use API tokens since they run outside the MCP context:
# Via plugin config (persistent)
# Set ROOTLY_API_TOKEN in the plugin's userConfig prompt
# Via env var (session-scoped)
export ROOTLY_API_TOKEN="your-token-here"Get a token from your Rootly dashboard under Settings > API Keys.
Commands
Stable (inline, MCP-native)
| Command | What It Does |
|---|---|
/rootly:setup | First-run configuration and connection check |
/rootly:my | Personal dashboard — your active incidents, action items, and upcoming on-call |
/rootly:status [service] | Service health overview — active incidents at a glance |
/rootly:oncall [team] | On-call dashboard with shift metrics |
/rootly:alert [short-id] | Triage a Rootly alert — events, group context, linked incident |
/rootly:lookup [name] | Look up a service, team, or catalog entity — owner, on-call, reliability |
/rootly:trend [scope] | 30-day reliability trend with prior-period comparison |
/rootly:brief [id] | Stakeholder summary for executives or customers |
/rootly:handoff [id] | Incident or on-call handoff documentation |
/rootly:ask [question] | Natural-language Q&A over your Rootly data |
/rootly:action [list|add|done] | Manage incident action items (write actions confirm before mutating) |
/rootly:swap [date] | Request someone cover one of your shifts (write, confirms first) |
/rootly:cover [team] | Offer to cover someone else's shift (write, confirms first) |
/rootly:announce [id] | Draft and post a stakeholder update on an incident (write, confirms first) |
Experimental (forked subagent — may not have MCP access in all contexts)
| Command | What It Does |
|---|---|
/rootly:respond [id] | Deep incident investigation via the incident-investigator agent |
/rootly:retro [id] | Post-incident retrospective via the retro-analyst agent |
/rootly:deploy-check | Pre-deploy risk analysis via the deploy-guardian agent |
The experimental skills delegate to forked subagents. In some Claude Code contexts the subagent doesn't inherit the plugin's MCP server; when that happens it stops and reports rather than falling back to bash/curl (which would leak the API token). For reliable coverage of those workflows, prefer
/rootly:brief,/rootly:status, and the inline alternatives above.
Natural Language Queries
/rootly:ask how many SEV-1 incidents did we have last month?
/rootly:ask which service has the most incidents this quarter?
/rootly:ask who's been on-call the most in the last 30 days?Deep Investigation Agents
When a slash command isn't enough, Claude automatically invokes specialized agents for deeper analysis:
| Agent | Triggered When | What It Does |
|---|---|---|
| Incident Investigator | You need root cause analysis beyond initial triage | Builds hypothesis trees, correlates alerts with code changes, traces causation chains |
| Deploy Guardian | Multi-service deployments with cross-team impact | Maps blast radius across dependent services, evaluates downstream risk, builds coordination checklists |
| Retro Analyst | You want to understand patterns across incidents | Clusters incidents by failure mode, calculates frequency trends, identifies systemic reliability issues |
Automatic Hooks
Two lightweight hooks run in the background -- they never block your workflow:
| Hook | When | What It Does |
|---|---|---|
| Token check | Session start | Validates your API token and nudges you to configure one if missing |
| Incident warning | Before git commit / git push | Warns if there's an active critical incident -- so you don't deploy into a fire |
Service Mapping
Map your repository to Rootly services by creating .claude/rootly-config.json:
{
"services": ["auth-service", "auth-worker"],
"team": "platform-team"
}/rootly:setup walks you through creating this. Without it, the plugin falls back to matching your git repo name against Rootly service names.
Advanced
<details> <summary><strong>Self-hosted Rootly</strong></summary>export ROOTLY_API_URL="https://rootly.internal.example.com"This overrides the REST API base URL used by hook scripts. Configure the MCP endpoint separately in .mcp.json.
Replace the HTTP transport in .mcp.json:
{
"mcpServers": {
"rootly": {
"command": "uvx",
"args": ["--from", "rootly-mcp-server", "rootly-mcp-server"],
"env": {
"ROOTLY_API_TOKEN": "${user_config.ROOTLY_API_TOKEN}"
}
}
}
}claude mcp add rootly --transport http https://mcp.rootly.com/mcpOAuth2 login will be triggered automatically on first use.
</details> <details> <summary><strong>Post-push deployment registration</strong></summary>An optional script (scripts/register-deploy.sh) can register deployments with Rootly after git push. It is not enabled by default -- see the script header for hook configuration.
Troubleshooting
| Problem | Fix |
|---|---|
| OAuth2 login doesn't open browser | Ensure your Rootly org has OAuth2 enabled. Try /mcp > disconnect > reconnect Rootly. |
| "No API token found" (hook scripts) | This only affects commit/push warnings. Set ROOTLY_API_TOKEN in plugin config or env var. MCP commands use OAuth2 instead. |
| MCP tools not responding | Disconnect and reconnect via /mcp, or reload with /reload-plugins. |
| OAuth2 consent shows limited permissions | Your org's OAuth2 configuration may need updating -- contact your Rootly admin. |
| Skills not appearing | Run /reload-plugins, then check the Installed tab in /plugin. |
| Hook scripts not running | Run chmod +x scripts/*.sh and ensure jq or python3 is available. |
Architecture
See ARCHITECTURE.md for the full technical design: MCP integration, hook system, agent orchestration, and data flow.
License
Apache 2.0 -- see LICENSE.
<p align=…