save-to-memory
Save current conversation to persistent memory. Use when user says "save this", "remember this", "store this for later", or wants to preserve context for future sessions.
When & Why to Use This Skill
This Claude skill enables persistent, encrypted memory storage for AI conversations, allowing users to save, organize, and recall critical context, decisions, and technical details across different sessions. It effectively solves the problem of context loss in LLM interactions by providing a structured local memory store categorized by specific domains like work, learning, and finance.
Use Cases
- Technical Documentation: Saving complex architectural decisions and refactoring logic to ensure continuity in future coding sessions.
- Project Continuity: Preserving the state of ongoing tasks, PR numbers, and specific implementation details to avoid re-explaining context to the AI.
- Knowledge Management: Storing key takeaways from research, tutorials, or learning sessions into dedicated 'Learning' or 'Work' spaces for long-term retrieval.
- Decision Tracking: Capturing the rationale behind business or technical conclusions reached during a conversation to serve as a persistent record.
| name | save-to-memory |
|---|---|
| description | Save current conversation to persistent memory. Use when user says "save this", "remember this", "store this for later", or wants to preserve context for future sessions. |
| allowed-tools | mcp__aiii__save_memory, mcp__aiii__list_spaces, mcp__aiii__create_space |
Save to Memory
Save the current conversation or specific content to aiii's local encrypted memory store for future recall.
When to Use
- User explicitly asks to save/remember something
- Important decision or conclusion was reached
- User wants to preserve context for future sessions
- After completing a significant task
Instructions
Analyze the conversation for key content worth saving:
- Main topics discussed
- Decisions made
- Code changes or implementations
- Important conclusions
Determine the appropriate space:
- Query available spaces using the MCP resource
aiii://spaces - Match content to existing spaces:
- Work: coding, tickets, PRs, technical work
- Health: medical, fitness, nutrition discussions
- Learning: tutorials, concepts learned
- Finance: budgets, investments, purchases
- If no suitable space exists, offer to create one
- Query available spaces using the MCP resource
Generate a concise title (under 50 characters):
- Use action-oriented language
- Include key identifiers (PR numbers, feature names)
- Examples: "Fixed auth bug in PR #234", "Vitamin D discussion"
Extract relevant tags:
- Technologies used (react, rust, python)
- Project names
- Key concepts
- People mentioned
Create a summary:
- 2-3 sentences capturing the essence
- Include outcomes and next steps if applicable
Call the save_memory tool:
{ "content": "Full conversation summary...", "title": "Concise title", "summary": "Brief 2-3 sentence summary", "space_id": "space_xxx", "source": "claude-code", "tags": ["tag1", "tag2"] }Confirm to user with:
- Space name where saved
- Title used
- Tags applied
Example Interaction
User: "Save this conversation about the auth refactoring"
Claude:
- Analyzes conversation → Found auth refactoring discussion
- Identifies space → Work
- Generates title → "Auth refactoring: JWT to session-based"
- Extracts tags → ["auth", "jwt", "refactoring", "security"]
- Creates summary → "Refactored authentication from JWT to session-based approach. Implemented HttpOnly cookies and CSRF protection. Next: add rate limiting."
- Saves to aiii
- Confirms: "Saved to Work space: 'Auth refactoring: JWT to session-based' with tags: auth, jwt, refactoring, security"