Skill

getting-started

Use when beginning any task, switching to an unfamiliar area of the codebase, or when context from previous sessions would help. Also use when resuming after context compaction.

npx claudepluginhub spawnbox-dev/claude-plugins --plugin orchestrator

Tool Access

This skill uses the workspace's default tool permissions.

Preview

<SUBAGENT-STOP>

SKILL.md

Similar Skills

ctx-remember

Recalls project context via ctx CLI: loads context packet, reads TASKS.md/DECISIONS.md, lists recent sessions, presents structured readback of last session, active tasks, recent decisions, next steps. For memory questions or session starts.

2 tools

ctx

cache-components

139.3k

Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.

cache-components

mcp-builder

124.2k

Guides building MCP servers enabling LLMs to interact with external services via tools. Covers best practices, TypeScript/Node (MCP SDK), Python (FastMCP).

9 files

anthropics-skills-13

Stats

Parent Repo Stars0

Parent Repo Forks0

Last CommitMar 18, 2026

Actions

View Source View Plugin View on GitHub View README

Help us improve

Share bugs, ideas, or general feedback.

Getting Started

You're entering a task and need context. Do this quickly and silently:

Call briefing to get the session briefing (open threads, recent decisions, last checkpoint)

If the task involves a specific domain or topic, invoke orchestrator:consult-concierge with a query like "What should I know about [topic]?" - the concierge searches conventions, decisions, anti-patterns, and architecture notes, then returns a curated summary. For simple lookups, call lookup directly instead.

For a complete picture of tracked work, call list_work_items (all statuses) and list_open_threads - these return everything without keyword search, so nothing is missed.

Scan the briefing for anything relevant to your current task

If the briefing shows a recovery checkpoint, honor it - that's where the last session left off

Do NOT dump the full briefing to the user. Internalize it and proceed with the task. Only mention relevant items (e.g., "I see there's an open thread about X that relates to this").

If you were dispatched as a subagent to execute a specific task, skip this skill. Subagents work from the context given to them, not from the full knowledge base. Do NOT respond to the user's first message until you have called `briefing` and reviewed the result. Do NOT skip this because the user's request seems simple or urgent. A 3-second briefing prevents 30 minutes of contradicting past decisions.

Getting Started

You're entering a task and need context. Do this quickly and silently:

Call briefing to get the session briefing (open threads, recent decisions, last checkpoint)
If the task involves a specific domain or topic, invoke orchestrator:consult-concierge with a query like "What should I know about [topic]?" - the concierge searches conventions, decisions, anti-patterns, and architecture notes, then returns a curated summary. For simple lookups, call lookup directly instead.
For a complete picture of tracked work, call list_work_items (all statuses) and list_open_threads - these return everything without keyword search, so nothing is missed.
Scan the briefing for anything relevant to your current task
If the briefing shows a recovery checkpoint, honor it - that's where the last session left off

Do NOT dump the full briefing to the user. Internalize it and proceed with the task. Only mention relevant items (e.g., "I see there's an open thread about X that relates to this").