Detects which coding-agent CLI sessions are waiting for human input across projects.
npx skills add https://github.com/vinzenz/workflows --skill agent-pager-vinzenz
Please help me install this Agent Skill. Skill: agent-pager Source: https://github.com/vinzenz/workflows/tree/main/.agents/skills/agent-pager Command: npx skills add https://github.com/vinzenz/workflows --skill agent-pager-vinzenz
waiting command shows only sessions blocked on a human, with a definitive empty state when none exist. - Peek without switching tabs: show <session> prints the last recorded event of a session so you can respond without changing windows. - Blocking wait loop: wait polls in the foreground until a session needs attention, honoring --timeout and --interval, and exits cleanly on expiry or Ctrl+C. - Use Case: You have four agent sessions running across three repositories; run agent-pager waiting to instantly see which one finished its turn and is waiting for your reply. ## Quick Start Ask the agent to run agent-pager and tell you which coding-agent sessions are currently waiting for your input.High-intent search queries and answers about installing and using this skill.
Run agent-pager with no arguments to list every live session with its state, or use agent-pager waiting to show only sessions blocked on human input. Filter by agent with --agent claude, codex, copilot, or opencode.
Use agent-pager wait, which blocks in the foreground until a session needs attention, then prints it and exits. It honors --timeout (default 1800s) and --interval, and exits 0 with a clear message if nothing needs you before expiry.
No, agent-pager opens every session store read-only and never writes, locks, or migrates data. A locked or corrupt store degrades to unavailable for that source while the rest of the view still renders.
It infers state from two signals: who spoke last (assistant message or completed tool result versus user message or in-flight tool call) and how long the store has been silent. Sessions silent over 24 hours are excluded entirely.
The heuristic can be wrong when a session is mirrored in two panes, when a killed process leaves a store looking waiting forever, or under clock skew. Unrecognized store formats are reported as unknown rather than guessed.
No, agent-pager only reads the last 24 hours of activity and never persists anything. For history, full transcripts, and search across past sessions, use the agent-log tool instead.
wiltaylorManage cloud agents via Warp's REST API and CLI.
cristoslcRestore terminal tab names, preferences, and context bookmarks across AI agent sessions.
moondrop-roFetch recent project changes from Git, Paperclip, Vercel, and Supabase adapters.
ma08Send contextual WhatsApp notifications to developers during coding agent workflows.