Monitor session context token levels and save checkpoints before compaction.
npx skills add https://github.com/velamints2/clawbot-lab --skill session-watchdog
Please help me install this Agent Skill. Skill: session-watchdog Source: https://github.com/velamints2/clawbot-lab/tree/main/skills/session-watchdog Command: npx skills add https://github.com/velamints2/clawbot-lab --skill session-watchdog
This skill helps maintain continuity in long conversations by tracking context token usage and proactively saving checkpoints before memory compaction.
Start a new session with the watchdog enabled to automatically monitor context and save checkpoints as your conversation grows.
High-intent search queries and answers about installing and using this skill.
To prevent data loss during session compaction, you can use an automated checkpoint protocol that monitors token usage and proactively saves memory files containing decisions and pending tasks before the context limit is reached.
You should create checkpoints during session starts, after every 30k tokens of conversation, before large tasks, and when users inquire about remaining context to maintain memory in long conversations.
The best way to monitor context token usage in extended design discussions is by applying a memory-based checkpoint protocol that tracks token levels and triggers user-facing alerts with threshold-based actions when approaching limits.
The checkpoint protocol works automatically by tracking tokens and triggering at specified levels, though it also responds to user inquiries about remaining context by offering options to continue, summarize, or pause.
When context is compacted, the memory file format preserves decisions, pending tasks, and important context details to ensure continuity and survive the context reset.
Yes, you can use session context monitoring for conversations that exceed standard token limits, as it tracks token usage and alerts you when approaching limits to offer options to continue, summarize, or pause.
guhans6Monitor Claude Code session token usage and estimate savings from session logs.
apexradiusRestore rules, plans, and memory after a context compaction event.
profmikegreeneDetect emerging anomalies and risk flags in live pilot metrics and data streams.
itsmostafaOptimizes LLM context windows via trimming, summarization, and external memory patterns.
