Watchdog Continuity Checks
Choose and word gentle wake, refocus, or checkpoint nudges for stalled agent workflows
- Repository
- mvschwarz/openrig
- Stars
- 8,553
- Licence
- Apache-2.0
- Files
- 1 file
What it does
This skill teaches an AI agent how to set up and reason about 'watchdog' interventions for multi-agent workflows (the 'rig' system). When an agent in a long-running workflow goes idle, drifts off course, or hits a phase boundary, the skill guides the agent to pick the lightest fitting nudge: a wake (restart motion), a refocus (re-center on the approved work), or an alignment checkpoint (a fuller pause to rebuild shared understanding). It also gives guidance on writing those nudge messages and avoiding over-frequent reminders.
Great for
- Developers configuring rig watchdog policies for long-running multi-agent workflows
- Teams trying to stop stalled or drifting agents without flooding them with reminders
- Anyone writing refocus or wake message text for agent orchestration
Skip it if
- Single-agent, short tasks where a scheduled nudge adds nothing; the skill itself says to prefer no-op for active owners
- Workflows with no approved plan or clear handoff, since refocus messages assume an approved workstream exists
- Environments without the openrig rig/daemon system; the guidance refers to tools and skills that only exist in that ecosystem
Try asking
- “Help me write a watchdog policy that wakes the consumer loop when a ready artifact appears in my rig.”
- “Which intervention level should I use if the reviewer agent has been silent for an hour but its handoff is still pending?”
- “Draft a refocus message for a seat that drifted into re-asking for approval on an already-approved workflow.”
- “Review my watchdog cadence: scan every 30 seconds, but how often should it actually wake agents?”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the Apache-2.0 licence and an attribution note.
watchdog.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip watchdog.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip watchdog.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give the agent your actual workflow state (queue items, which seat owns what) so it can choose between wake and refocus based on evidence rather than guessing.
- Start with one workflow-level watchdog and add targeted exceptions only when needed, as the skill recommends, to avoid nag loops.
- Keep scan frequency and wake frequency separate; scanning often is cheap, waking often pollutes the workstream.
What's inside
- SKILL.md
Redistributed unmodified from mvschwarz/openrig under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.