🚨 incidents
Check whether a vendor is having an incident before you debug your own code — reads the status pages of the services you depend on through the incidents CLI.
The incidents plugin wraps the
@kud/incidents-cli CLI. It
ships a skill only, with no MCP server. When a third-party call is failing, the
skill runs the CLI, reads what each vendor is declaring on its own status page,
and tells you whether the fault is likely theirs before you go digging in your
own code.
What it bundles
| Skill | Trigger | What it does |
|---|---|---|
incidents-check | An external API or SaaS call failing, timeouts or 5xx from a vendor, "is GitHub/AWS down?" | Runs incidents list --json, reads it worst first, and treats an unreadable page as not green, never as fine |
Install
/plugin install incidents@kud --scope userConfiguration
Install the CLI the skill drives (Node 22 or later):
npm i -g @kud/incidents-cliThe global install is optional: without it, the plugin ships its own incidents that runs the CLI through npx.
The services come from your own config, ~/.config/incidents/config.yml plus
any config.d/*.yml. With no config it watches github, npm, cloudflare
and claude. The skill only reads: it never edits your config.
A green row is what the vendor has declared, not proof that your calls work.
Links
- npm:
@kud/incidents-cli - GitHub: kud/incidents-cli