Menu
Identify unlabeled GitHub issues that may belong to a specific team
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Monitor CI checks after pushing, detect flaky vs legit failures, and auto-fix
End-of-day counterpart to wake-me-up. Reads today's morning briefing at ~/dev/ai/notes/wake-me-up/{date}.md, asks the user what they shipped or resolved, drafts a Slack wrap-up post in the user's voice, shows the draft for explicit confirmation, then posts it on the user's behalf via the Slack MCP. Appends the posted summary back to the briefing file so the file is a full day record. Use when the user says "/closing-time", "wrap the day", "end of day post", or wants help summarizing what they did.
Commit staged/unstaged changes with a well-crafted commit message.
Create or update a GitHub PR with automatic template detection and filling
Plan, implement, and iteratively review a task end-to-end using Claude + Copilot reviewers in a linear flow.
Investigate what it would take to upgrade a library from its currently pinned version to a target version (or "latest"). Produces a migration assessment — version delta, breaking changes that apply to us, impacted files, estimated blast radius, suggested rollout — and STOPS there. Does not perform the upgrade unless the user explicitly asks. Use when the user says "investigate updating <pkg> from X to Y", "investigate updating <pkg> to latest", "we are behind on <pkg>", or "let's update <pkg> to version N".
| name | triage-issues |
| description | Identify unlabeled GitHub issues that may belong to a specific team |
Identify unlabeled GitHub issues that may belong to a specific team.
Example invocations:
/triage-issues → defaults (14 days, 30 issues, feature-flags team)/triage-issues 7 → last 7 days/triage-issues 7 50 → last 7 days, up to 50 issues/triage-issues for web-analytics → triage for web-analytics team/triage-issues last 7 days for product-analytics → natural language works tooExtract from the user's input (or use defaults):
days = number of days to look back (default: 14)limit = max issues to fetch (default: 30)team = team identifier (default: feature-flags)Supported team identifiers:
feature-flags or ff → Feature Flags teamweb-analytics or wa, product-analytics or pa, etc.)If an unsupported team is requested, inform the user which teams are available.
Fetch unlabeled issues from PostHog/posthog using gh:
gh issue list --repo PostHog/posthog --state open --limit {limit} --json number,title,body,labels,createdAt,url --search "created:>=$(date -v-{days}d +%Y-%m-%d) {exclusion_labels}"
The {exclusion_labels} vary by team. For feature-flags:
-label:team/feature-flags -label:feature/feature-flags -label:feature/cohorts -label:feature/early-access
Note: The date -v-Nd syntax is macOS-specific. On Linux, use date -d "N days ago".
Use the Task tool to spawn the appropriate triage subagent:
feature-flags: Use subagent triage-feature-flagsPass the fetched issues to the subagent for analysis. The subagent will:
Show the user:
Ask the user which issues to label:
Apply labels using:
gh issue edit --repo PostHog/posthog {number} --add-label "{labels}"
To add support for a new team:
ai/agents/triage-{team-name}.md