Quick Start

This guide walks you through building PRX-Memory, running the daemon, and performing your first store and recall operations.

1. Build the Daemon

git clone https://github.com/openprx/prx-memory.git
cd prx-memory
cargo build -p prx-memory-mcp --bin prx-memoryd

2. Start the Server

Option A: stdio Transport

For direct MCP client integration:

PRX_MEMORYD_TRANSPORT=stdio \
PRX_MEMORY_DB=./data/memory-db.json \
./target/debug/prx-memoryd

Option B: HTTP Transport

For network access with health checks and metrics:

PRX_MEMORYD_TRANSPORT=http \
PRX_MEMORY_HTTP_ADDR=127.0.0.1:8787 \
PRX_MEMORY_DB=./data/memory-db.json \
./target/debug/prx-memoryd

Verify the server is running:

curl -sS http://127.0.0.1:8787/health

3. Configure Your MCP Client

Add PRX-Memory to your MCP client configuration. For example, in Claude Code or Codex:

{
  "mcpServers": {
    "prx_memory": {
      "command": "/path/to/prx-memory/target/release/prx-memoryd",
      "env": {
        "PRX_MEMORYD_TRANSPORT": "stdio",
        "PRX_MEMORY_BACKEND": "json",
        "PRX_MEMORY_DB": "/path/to/prx-memory/data/memory-db.json"
      }
    }
  }
}

TIP

Replace /path/to/prx-memory with the actual path where you cloned the repository.

4. Store a Memory

Send a memory_store tool call through your MCP client or directly via JSON-RPC:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "memory_store",
    "arguments": {
      "text": "Always use parameterized queries for SQL to prevent injection attacks",
      "scope": "global",
      "tags": ["security", "sql", "best-practice"]
    }
  }
}

5. Recall Memories

Retrieve relevant memories using memory_recall:

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "tools/call",
  "params": {
    "name": "memory_recall",
    "arguments": {
      "query": "SQL security best practices",
      "scope": "global",
      "limit": 5
    }
  }
}

The system returns memories ranked by relevance using a combination of lexical matching, importance scoring, and recency.

6. Enable Semantic Search (Optional)

For vector-based semantic recall, configure an embedding provider:

PRX_EMBED_PROVIDER=jina \
PRX_EMBED_API_KEY=your_jina_api_key \
PRX_EMBED_MODEL=jina-embeddings-v3 \
PRX_MEMORYD_TRANSPORT=stdio \
PRX_MEMORY_DB=./data/memory-db.json \
./target/debug/prx-memoryd

With embeddings enabled, recall queries use vector similarity in addition to lexical matching, significantly improving retrieval quality for natural language queries.

7. Enable Reranking (Optional)

Add a reranker to further improve retrieval precision:

PRX_EMBED_PROVIDER=jina \
PRX_EMBED_API_KEY=your_embed_key \
PRX_EMBED_MODEL=jina-embeddings-v3 \
PRX_RERANK_PROVIDER=cohere \
PRX_RERANK_API_KEY=your_cohere_key \
PRX_RERANK_MODEL=rerank-v3.5 \
PRX_MEMORYD_TRANSPORT=stdio \
PRX_MEMORY_DB=./data/memory-db.json \
./target/debug/prx-memoryd

Available MCP Tools

Tool	Description
memory_store	Store a new memory entry
memory_recall	Recall memories by query
memory_update	Update an existing memory
memory_forget	Delete a memory entry
memory_export	Export all memories
memory_import	Import memories from export
memory_migrate	Migrate storage format
memory_reembed	Re-embed memories with new model
memory_compact	Compact and optimize storage
memory_evolve	Evolve memory with holdout validation
memory_skill_manifest	Discover available skills

Next Steps

• Embedding Engine -- Explore embedding providers and batch processing
• Reranking -- Configure second-stage reranking
• Storage Backends -- Choose between JSON and SQLite storage
• Configuration Reference -- All environment variables