Where are my research checkpoints stored?

All checkpoints are saved as YAML files in the `~/.sage/checkpoints/` directory on your local machine, making them accessible across different projects.

Can I restore a checkpoint in a new session?

Yes, the skill is designed to inject saved context into new sessions, allowing you to resume complex research exactly where you or the AI left off.

When does this skill automatically trigger?

It triggers when the AI detects synthesis points (e.g., 'In summary'), topic shifts, validated hypotheses, or when you explicitly ask to 'save research'.

How does it help with long context windows?

By compressing research into decisions and conclusions rather than raw chat history, it allows the AI to 'remember' the core findings of an investigation without hitting token limits.

What is a research checkpoint?

A research checkpoint is a structured snapshot of the AI's current understanding, including its thesis, sources, and open questions, saved locally to ensure continuity.

Sage Research Checkpoint

Name: Sage Research Checkpoint
Author: b17z

byb17z

•

生産性とワークフロー

Preserves AI research context and decision states through semantic checkpoints to prevent information loss during complex investigations.

概要

Sage Research Checkpoint empowers Claude to proactively save and restore research states using a structured format that captures theses, evidence, and open questions. By creating semantic checkpoints stored locally, it facilitates seamless session handoffs and ensures that deep technical investigations remain coherent even across large context windows. The skill intelligently identifies key moments—such as discovering constraints, reaching conclusions, or shifting topics—to automatically trigger state preservation, making it an essential tool for high-stakes analysis and multi-step problem solving.

主な機能

Proactive semantic state preservation using a structured YAML format
Efficient restoration of research context for seamless session handoffs
Context-aware autosaving triggered by synthesis, topic shifts, or user validation
Local storage in a global directory for persistent cross-project recall
Detailed tracking of research tensions, confidence levels, and source relations
1 GitHub stars

ユースケース

Collaborative research handoffs between different AI sessions or human-AI teams
Complex decision-making processes requiring a clear audit trail of evidence and synthesis
Long-running technical investigations where context depth might exceed token limits

概要

主な機能

Proactive semantic state preservation using a structured YAML format
Efficient restoration of research context for seamless session handoffs
Context-aware autosaving triggered by synthesis, topic shifts, or user validation
Local storage in a global directory for persistent cross-project recall
Detailed tracking of research tensions, confidence levels, and source relations
1 GitHub stars

ユースケース

Collaborative research handoffs between different AI sessions or human-AI teams
Complex decision-making processes requiring a clear audit trail of evidence and synthesis
Long-running technical investigations where context depth might exceed token limits