---

name: memory-read description: Search and retrieve past decisions from Mother-Harness long-term memory

Memory Read Skill

Retrieves relevant past decisions and learnings from persistent memory.

When to Use

Use this skill when you want to:

• Recall past decisions on a similar topic
• Find rationale for previous choices
• Search for all decisions in a project
• Look up decisions by tag

Input Format

Provide a search query:

CopyQuery: [Natural language query or keywords]
Project: [Filter by project - optional]
Tags: [Filter by tags - optional]
Limit: [Max results, default 5 - optional]

Examples

Semantic Search

CopyQuery: database selection decisions
Limit: 3

Project-Scoped

CopyQuery: authentication approach
Project: payment-platform

Tag Filter

CopyQuery: recent infrastructure decisions
Tags: architecture, infrastructure

What Happens

1. Query is sent to the n8n memory-read webhook
2. Semantic search finds relevant vectors in Qdrant (using Ollama embeddings)
3. Results are returned with similarity scores

Response Format

Copy{
  "results": [
    {
      "decision": "Use PostgreSQL over MongoDB",
      "rationale": "Strong consistency...",
      "project": "payment-platform",
      "tags": ["database"],
      "timestamp": "2024-01-15T10:30:00Z",
      "score": 0.92
    }
  ],
  "total": 5,
  "query": "database selection decisions"
}

Implementation

Call the n8n webhook at: http://localhost:5678/webhook/memory-read

Copycurl -X POST http://localhost:5678/webhook/memory-read \
  -H "Content-Type: application/json" \
  -d '{
    "query": "database selection decisions",
    "limit": 5
  }'