How do I install the Mo Watchdog skill?

Run npx skillmds@latest add shkarupa-alex/mo-watchdog in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Mo Watchdog safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Mo Watchdog?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Mo Watchdog free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under MIT.

Who published Mo Watchdog?

shkarupa-alex (@shkarupa-alex) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Mo Watchdog by shkarupa-alex — Agent Skill

Name: Mo Watchdog
Author: shkarupa-alex

Mo Watchdog

Observe one target or scan reachable Orca sessions for limits, overload, failures, questions, work, and completion. Use only when the user explicitly names Meta-O watchdog or mo-watchdog; nudge only an explicitly authorized exact target.

shkarupa-alex 58d6ebd 4 files · 31.1 KB Updated Sep 17, 2026

File contents

Watch backend sessions

Read Watchdog behavior completely. Start only after the user explicitly names Meta-O watchdog or mo-watchdog; a generic monitoring question does not activate this skill.

Use scripts/mo-watchdog.sh target --backend orca --session <id> for one session or scripts/mo-watchdog.sh scan for all reachable Orca sessions. Observation is read-only. An explicit nudge additionally requires --nudge <message> and exact target authorization; the script re-reads native state and suppresses the nudge when that state changed. It reserves a bounded private digest before delivery and suppresses the same message, an ambiguous attempt, or a saturated unchanged state across later invocations. Nudges return after native delivery; agent completion is observed separately.

Treat malformed or stale native state as a typed unsafe observation: report it, fail closed, and do not nudge or guess the target state.

Do not inspect tracked project content or private provider state. Report the native locator, classified state and action. Pattern misses are refined from observed failures rather than hidden behind a support claim.

Frequently asked questions