diverga-memory
Diverga Memory System v7.0 - Context-persistent research support with checkpoint auto-trigger and cross-session continuity. 기억, 맥락, 세션, 체크포인트
Profile →npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill memory --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Diverga Memory System v7.0 provides context-persistent research support with checkpoint auto-trigger and cross-session continuity. It supports loading and displaying project context across sessions, auto-injecting full research context when subagents are used, and managing a multi-layer context system with a 3-layer structure, a decision audit trail, and automated research documentation.
How it works
- Context loading uses keyword triggers to display the current project state, loading project-state.yaml and showing stage, progress, pending checkpoints, and next actions.
- Task Interceptor injects full research context into agent prompts when a subagent_type with the diverga: prefix is called. It reads project-state.yaml and checkpoints.yaml, then injects a structured research_context into the prompt and wraps the checkpoint validation for execution.
- CLI Layer allows explicit prompts via /diverga:memory context with flags like --verbose, --archive, --decisions, --checkpoints, and --format to output the state in json|yaml|text.
- Checkpoint system defines levels (REQUIRED, RECOMMENDED, OPTIONAL) with a standard set of checkpoints across Foundation, Design, Planning, Execution, and Validation stages. Checkpoints are tracked in decision-log.yaml and checkpoint status is visible in status outputs.
- Decision Audit Trail records immutable, versioned decisions with timestamps, rationale, and prior decisions. Amendments create new entries (e.g., DEV_002_A1) linked to original (DEV_002).
When to use it
Use when starting a research project to initialize a structured .research directory, define checkpoints, and maintain cross-session continuity. Use the status or context commands to review current state and upcoming actions; use the decision commands to log and amend pivotal decisions.
What it can touch
- Commands: "/diverga:memory init", "/diverga:memory status", "/diverga:memory context", "/diverga:memory decision list", "/diverga:memory migrate".
- Tools: CLAUDE-CODE (as declared) for interacting with the system.
- Files referenced: .research/project-state.yaml, .research/checkpoints.yaml, .research/decision-log.yaml, .research/priority-context.md (for updated priority context), and related session and archive records.
Caveats
- The skill’s behavior relies on existence of .research directory and YAML files described (project-state.yaml, checkpoints.yaml, decision-log.yaml).
- Checkpoint completion and decision amendments are tracked with timestamps and versioning; ensure correct entries to maintain audit trail.
- Output formats and injected context depend on proper keywords and subagent_type syntax; misframing could affect prompt injection.
# Diverga Memory System v7.0 ## Overview Human-centered research context persistence with: - 3-Layer Context System - Checkpoint Auto-Trigger - Cross-Session Continuity - Decision Audit Trail - Research Documentation Automation ## Quick Reference ### Context Loading Keywords **English**: "my research", "research status", "where was I", "continue research", "what stage" **Korean**: "내 연구", "연구 진행",
What does the diverga-memory skill do?
Diverga Memory System v7.0 - Context-persistent research support with checkpoint auto-trigger and cross-session continuity. 기억, 맥락, 세션, 체크포인트
How do I install it?
Run `npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill memory --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From brycewang-stanford/Auto-Empirical-Research-Skills, a repository with 3,244 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.