agents

This skill should be used when the user asks about "how to store data in a plugin", "what file format to use for plugin storage", "YAML vs JSON in a plugin", "how to keep task files in sync", "streaming storage", "frontmatter metadata", "data storage for hooks", "writing files in MCP servers", "concurrent writes in plugin storage", "file format selection", "storing plugin state", or "persistent data in Claude Code plugins".

Review a GitHub pull request and post inline feedback via the GitHub review system. Used by the run-agent composite action; the body below is the prompt that drives claude-code-action.

How to use tmux effectively when working with Claude Code agent teams — session hierarchy, attach/detach, scripting teammate panes, common pitfalls.

How to author agent files (frontmatter + system prompt body) for the agent-teams project — file structure, naming, frontmatter fields, prompt conventions.

Use this skill when you need to fetch public Reddit content (posts, comments, subreddits, user activity) without authentication. Trigger when: - User asks about Reddit posts, comments, or subreddit content - Research requires public Reddit data - Fetching specific Reddit posts by URL

Use at every task-lifecycle decision point (before TaskCreate, before TaskUpdate→in_progress, after TaskUpdate→completed) when you need help applying the task-management doctrine — atomicity check, breakdown planning, status-transition validity, or follow-up capture. Forks into an isolated context so the parent's window stays lean. Returns a ≤5-sentence imperative instruction the parent should execute. Trigger phrases — "is this task atomic?", "should I break this down?", "what task should I start next?", "I just completed a task, what now?", "plan my next move on the task list", "audit my task list discipline".

Use when the handler asks to fold user-scope settings into the canonical repo-scope file, when an agent restart wiped out plugin enablement / marketplace registrations that need to be recovered, or when audit shows drift between `$CLAUDE_CONFIG_DIR/settings*.json` and `$AGENT_REPO/.claude/settings.json`. Trigger phrases — "merge my settings", "fold user-scope into repo-scope", "settings-merge", "recover the settings drift", "what plugins are enabled in my user-scope but not committed?". Requires `$AGENT_REPO/.claude/settings.json` to already exist as the target.

Use when the agent-utils settings-write-guard PostToolUse hook prints a 2-line WARN about a settings.json write happening outside $AGENT_REPO/.claude/settings.json, OR before writing/editing any ~/.claude/settings*.json (user-scope) file. Trigger phrases — "the hook is firing on my settings write", "I got a wiped-on-agent-restart warning", "where should I redirect this settings edit?", "/plugin install added something — does it persist?". See `what-survives-agent-restart` for the doctrine this enforces.

Use when deciding WHERE to put persistent configuration in a Claude Code agent — plugin enablement, marketplace registration, hook wiring, custom commands, or any settings that need to outlive a restart. Trigger phrases — "where should this setting go?", "why did my plugin disappear after restart?", "is ~/.claude/settings.json the right place for X?", "what gets wiped on restart?", "should I put this in repo-scope or user-scope settings?".

Use when an active conversation has gone idle and the agent is blocked waiting for answers or actions from people. Decides whether enough idle time has passed (default 30m since the last response from the blocker) to send a single, well-formatted reminder. Returns either a "send reminder now" instruction with the exact message text, or "stay quiet" with the reason. Trigger phrases — "the convo's been idle for a while, should I ping?", "I've been waiting on X for N minutes, should I follow up?", "is it time to remind Nate about the open question?".

Cron-delegate skill for the 5-minute self-poll. Invoked every 5 minutes by cron — checks (a) in_progress tasks, (b) MONITORING(...) pending tasks, (c) AGENT(...) pending tasks, and any background shells; applies the self-correction discipline ladder (1st/2nd/3rd-fire); delegates to active-convo-goes-idle when the conversation has been idle past 30m. Trigger phrases — "5m self-poll fired", "cron tick", "check what I'm waiting on", "Skill(idle-5-min)".