ci-triage
Diagnose a failing GitHub Actions run — find the first real error in the logs, tell a flake apart from a genuine failure, and identify the commit that broke…
Collect and submit feedback about Superset (bug reports, feature requests, or general feedback) privately to the Superset team or as a public GitHub issue. Use when the user wants to report a Superset bug, request a feature, or send feedback about Superset.
$ npx -y skills add superset-sh/superset --skill feedback --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/feedbackContext preview
The summary Claude sees to decide when to auto-load this skill.
Collect and submit feedback about Superset (bug reports, feature requests, or general feedback) privately to the Superset team or as a public GitHub issue. Use when the user wants to report a Superset bug, request a feature, or send feedback about Superset.
name: feedback description: Collect and submit feedback about Superset (bug reports, feature requests, or general feedback) privately to the Superset team or as a public GitHub issue. Use when the user wants to report a Superset bug, request a feature, or send feedback about Superset. argument-hint: describe the bug, request, or feedback allowed-tools: Bash(superset:*) Bash(gh:*) Bash(uname:*)
Turn the user's feedback about Superset into a short, scannable report and submit it where they choose. Treat whatever they wrote after the command as the seed.
The reader is a Superset engineer triaging dozens of reports. They should get the point from the title, the full picture from the summary bullets, and only read further when they need detail. Cut everything that doesn't help them reproduce or decide.
Every private submission CCs the reporter, so they already have a copy in their inbox with a subject starting `[Feedback: ...]`. A reply to that email lands on the same thread; a new submission opens a new one and the team has to reconcile them by hand.
If the seed reads like a continuation ("follow-up", "again", "still happening", "update on", "retraction", "root cause", "I already sent", "third time"), or the user filed something on this topic earlier in the session, ask whether they already sent a report about it. If they did, tell them to reply to that email with the new information and stop here. Only file fresh when they confirm it's a different issue.
Never title a report "Follow-up:", "ROOT CAUSE:", or "RETRACTION"; those belong in a reply.
Run in parallel; skip anything that fails:
Don't include repository contents, terminal output, or logs unless the user explicitly agrees when asked; they can contain private paths and code.
Many reports describe bugs already fixed in a newer release. Desktop, CLI, and host service share one version number, so compare what's running against the latest release:
If anything is behind, say so before drafting and ask whether they want to update first (the desktop app updates itself from its menu; a standalone CLI with `superset update`). If they'd rather file now, note the versions in the report.
Symptoms like "host not registered", "hosts list is empty", "automations create fails", "host shows offline", "terminal won't attach", or "workspace missing from the sidebar" are the most common reports and usually have a known cause. Before drafting, run the doctor skill's snapshot (`superset status`, `superset auth whoami`, `superset hosts list`) and apply its signature table. If that fixes it, there's nothing to report. If it doesn't, keep the `superset status --json` output for the report; it's what the team asks for first.
Classify as **bug**, **feature request**, or **general feedback** from their words; ask only if genuinely ambiguous.
**One report per root issue, one request per ask.** Two symptoms of the same failure are one bug. Three unrelated wishes are three feature requests; offer to split them, then file each separately. Don't file a second report for a different symptom of something the user just filed.
`<Surface>: <symptom or ask>`, under 72 characters, no trailing period.
Good: `Terminal: pane goes blank after waking from sleep`, `Sidebar: let me pin an automation to the top`. Bad: `Bug with terminal`, `Terminal rendering broken because xterm loses WebGL context on resume`.
Plain text that also reads well as markdown: section labels on their own line, `-` bullets, numbered steps. No headings, bold, or tables; the private path is delivered as a plain-text email. Keep the user's own words where they're precise, tighten where they ramble. A one-line seed still gets Summary bullets.
Every report starts with a **Summary** of 2-4 bullets that stand alone. Someone reading only those bullets should know what's wrong (or wanted), how bad it is, and how often it happens. Summary bullets are observations, never hypotheses.
Bug:
Summary - Terminal pane goes blank after the Mac wakes from sleep - Happens every time; reloading the window fixes it - Started in 1.21.0, didn't happen in 1.20.x Steps to reproduce 1. Open a workspace with one terminal running 2. Close the lid for a minute, then open it 3. Click into the t
Superset is an agentic IDE to orchestrate 100+ coding agents in parallel. Run any agent with your own subscription.
Diagnose a failing GitHub Actions run — find the first real error in the logs, tell a flake apart from a genuine failure, and identify the commit that broke…
Triage a GitHub issue into something actionable — reproduce the claim, find duplicates, judge severity, and apply labels, milestone, and assignee. Use when the…
Find and merge duplicate Linear issues — group reports of the same underlying bug, pick the survivor, and move the evidence across. Use when the backlog has…
Turn a rough report into a Linear issue someone can pick up — reproduce the claim, check for duplicates, and fill in team, priority, and labels. Use when the…
Draft a Linear project status update from what actually moved — progress, risks, and the one decision that needs making. Use when someone asks for a project…
Personalized audit that teaches the advanced Superset features the user isn't using yet (automations, parallel agents, tasks, multi-host, terminal remote…