work-summary_skill

Overview

This skill generates an insight-driven project dashboard that summarizes active work, blockers, stalled items, and recommended next actions. It interprets task and container signals rather than just listing items, surfacing anomalies, bottlenecks, and concrete remediation steps. Use it to get a concise, manager-style view of project health and immediate priorities.

How this skill works

The skill queries three data sources in parallel: a root-level overview for container health, the work context for active/blocked/stalled items with ancestor chains, and a short list of prioritized next actions. It classifies root items (active containers, empty containers, completed containers, anomalies) and analyzes patterns like single-threading, blocked dependencies, missing notes, and recommendation quality. The output is a compact dashboard with Active Work, Needs Attention, Up Next, Container Health, Empty, and Done sections, plus short agent observations where they add value.

When to use it

• Ask for a high-level status: “project status”, “work summary”, or “show me the dashboard”.
• You want actionable interpretation of task metadata rather than a raw list.
• Before planning a sprint or standup to identify blockers and clear next actions.
• When you suspect work is single-threaded or a container was closed prematurely.
• To identify planning gaps (empty containers) and verify container momentum.

Best practices

• Scope the query with a project UUID or clear text to focus the dashboard on a single area.
• Run the dashboard early in planning to expose missing requirements or done-criteria notes.
• Treat agent observations as prompts for human follow-up: they flag where to ask for missing context or escalate blockers.
• Use the Up Next list to assign immediate work and look for parallelization opportunities.
• Rely on Container Health signals to decide whether to start, pause, or re-prioritize whole streams.

Example use cases

• Daily standup summary showing 3 active items, 1 blocker, and two recommended next actions.
• Find stalled work: surface items missing requirements or done-criteria and suggest the precise missing notes.
• Scope-specific status: run against a feature UUID to get only that feature’s active work and recommendations.
• Pre-release check: identify review-role items and container completion ratios to evaluate release readiness.
• Backlog health audit: list empty containers that represent planning gaps to trigger backlog grooming.