memory-management

name	memory-management
description	Search and manage Aliceloop's memory and conversation recall. Use when the user asks about past conversations, personal facts, preferences, or anything that requires recalling information ("do you know my...", "we talked about before...", "do you remember...", "help me find what we said about..."). Also used to store new durable memories and search through archived chat threads.
allowed-tools	["bash","read","write"]

Memory Management Skill

Aliceloop has a built-in recall system with two layers:

• durable memory for stable facts and preferences
• conversation history search for past thread wording

Use the aliceloop CLI to interact with it.

Tools

# List memories
aliceloop memory list [limit]

# Semantic search
aliceloop memory search "<query>"

# Search conversation history
aliceloop memory grep "<query>"

# Force transcript archive resync
aliceloop memory archive

# Add a memory
aliceloop memory add "<content>"

# Delete a memory
aliceloop memory delete <id>

When to Use

• User asks anything about the past ("do you know what I like", "what did we discuss before", "what was that plan we talked about last time") → Search memory AND grep threads
• User says "remember this" → aliceloop memory add "..."
• User asks "do you remember..." → aliceloop memory search "..." + aliceloop memory grep "..."
• User says "forget about..." → Search and delete matching memories
• Time-sensitive info (projects, deadlines) → Store with appropriate context

Search Strategy

When the user asks about past information, always try both layers when needed:

1. aliceloop memory search "<query>" for semantic recall of stable facts
2. aliceloop memory grep "<keyword>" for keyword search in conversation history

If one layer is not enough, use the other. They complement each other. On routed recall turns, Aliceloop may load the retrieved memory results into the conversation context before answering.

Two-Layer Recall

1. Vector Memory (aliceloop memory search) — semantic search, finds conceptually related memories
2. Conversation History (aliceloop memory grep) — keyword search, finds exact words and phrases in past threads

Use semantic search when the user asks vague recall questions. Use grep when you need specific terms, names, wording, or code snippets.

Conversation History Search

Aliceloop automatically exports project-backed thread transcripts as markdown files. memory grep prioritizes those conversation archives in threads/ before falling back to live runtime history.

# Keyword search through archived conversations
aliceloop memory grep <keyword>

# Force re-export of thread archives now
aliceloop memory archive

Conversation archives are stored in the project's threads/ directory as markdown files with frontmatter (threadId, title, createdAt, updatedAt, model, messageCount). They are auto-exported when sessions change and can be resynced explicitly with memory archive.

Tips

• Always confirm what you stored/deleted with the user
• Use aliceloop memory search to find related memories before adding duplicates
• Do not claim a memory exists unless retrieval actually found it
• Use aliceloop memory grep to search conversation archives when semantic recall is not enough
• Conversation archives live in the project's threads/ directory and are auto-exported as markdown files