Setup & Installation
openclaw skills install @xbillwatsonx/session-watchdog
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
npx clawhub install session-watchdog
What This Skill Does
Monitors token usage in active Claude sessions and triggers checkpoint saves to memory files before context compaction occurs. Tracks four threshold levels from safe to full (0–197k+ tokens) and prompts the user to continue, restart, or pause when nearing limits.
Saves context state before compaction happens automatically, giving users a choice instead of silently losing conversation history.
When to use it
- Checking remaining context before starting a large file edit
- Saving pending tasks before a long conversation compacts
- Answering 'how much context is left?' mid-session
- Preserving decisions and configurations across session boundaries
- Starting a new session with a recap of unresolved issues
Example Workflow
Here's how your AI assistant might use this skill in practice.
INPUT
User asks: How much context do we have left?
AGENT
- 1Run session_status to retrieve contextTokens value
- 2Compare tokens against defined thresholds (Safe / Warning / Critical / Full)
- 3If above 160k, read the current memory/YYYY-MM-DD.md file
- 4Append checkpoint with decisions, pending tasks, and modified file paths
- 5Alert user with current percentage and ask whether to continue, restart, or pause
OUTPUT
Context is at ~82% (160k tokens). Checkpoint saved to memory/2026-03-19.md with pending tasks and recent decisions. User prompted to choose next step.