Persist timestamped decisions, evidence, and resume points for agent tasks.
npx skills add https://github.com/thistleknot/skills --skill continuity-log
Please help me install this Agent Skill. Skill: continuity-log Source: https://github.com/thistleknot/skills/tree/main/continuity-log Command: npx skills add https://github.com/thistleknot/skills --skill continuity-log
Continuity-log solves the problem of losing high-value decisions and resume context during long, multi-turn, or compaction-heavy agent runs, which otherwise leads to repeated analysis and wasted cycles.
Write a timestamped continuity entry capturing the objective, decision, evidence, rejected alternatives, and the exact resume point to the most appropriate continuity location for your run.
High-intent search queries and answers about installing and using this skill.
To preserve agent decisions during context compaction, persist load-bearing outcomes, evidence, and rejected alternatives into a durable continuity log instead of storing raw chain-of-thought. This prevents future turns from repeating prior analysis and wasting cycles.
The best way to resume a long coding task after an interruption is to record a clear resume point in a timestamped continuity log. This allows the next agent run to quickly reconstruct the operational path without re-running the same reasoning.
Decision logging for multi-agent orchestration works by capturing compact-safe decision packets that store outcomes, blockers, and open questions. This ensures subsequent agents can read the log before resuming and avoid duplicating previous analytical work.
Yes, you can use continuity logging for research tasks hitting context limits. It records where the next turn should continue after compaction, preserving what is known, what is uncertain, and what must be revisited across long-running research sessions.
When persisting working memory for agent tasks, you should strictly avoid storing raw hidden chain-of-thought and unnecessary sensitive data. Focus solely on persisting durable decisions, evidence, and resume points to maintain compact-safe logs.
YuvaseeDefine tasks interactively through Q&A and generate timestamped Markdown documentation.
richardmbaileyMaintain structured markdown diaries of decisions, findings, and next actions.
Owl-ListenerCoordinate agent-to-human handoffs with triggers, sources, destinations, payload, and acknowledgment.
HosungYouPersist research context, decision audit trails, and checkpoints across sessions.
