@opencode-manager/memory

@opencode-manager/memory

Memory management plugin for OpenCode that enables semantic search and persistent storage of project knowledge.

Features

• Semantic Memory Search - Store and retrieve project memories using vector embeddings
• Multiple Memory Scopes - Categorize memories as convention, decision, or context
• Automatic Deduplication - Prevents duplicate memories from being stored
• Session Context - Tracks conversation history and provides relevant memories during chats
• Tool Usage Tracking - Records tool executions for context-aware memory retrieval

Tools

Tool	Description
memory-read	Search and retrieve project memories with semantic search
memory-write	Store a new project memory
memory-edit	Update an existing project memory
memory-delete	Delete a project memory by ID
memory-health	Health check or full reindex of the memory store

Installation

Install the package from npm:

npm install @opencode-manager/memory
# or
pnpm add @opencode-manager/memory

During installation, the local embedding model (all-MiniLM-L6-v2) is downloaded automatically via the postinstall script.

Then configure opencode to load the plugin. In your opencode.json:

{
  "plugin": "@opencode-manager/memory"
}

Configuration

On first run, the plugin automatically copies the bundled config to your data directory:

• Path: ~/.local/share/opencode/memory/config.json
• Falls back to: $XDG_DATA_HOME/opencode/memory/config.json

You can edit this file to customize settings. The file is created only if it doesn't already exist.

{
  "embedding": {
    "provider": "local",
    "model": "all-MiniLM-L6-v2",
    "dimensions": 384,
    "baseUrl": "",
    "apiKey": ""
  },
  "dataDir": "~/.local/share/opencode/memory",
  "dedupThreshold": 0.9,
  "logging": {
    "enabled": false,
    "file": "~/.local/share/opencode/memory/logs/memory.log"
  }
}

Options

Embedding

• embedding.provider - Embedding provider: "local", "openai", or "voyage"
• embedding.model - Model name
  • local: "all-MiniLM-L6-v2" (384d) or "nomic-embed-text-v1.5" (768d)
  • openai: "text-embedding-3-small" (1536d), "text-embedding-3-large" (3072d), or "text-embedding-ada-002" (1536d)
  • voyage: "voyage-code-3" (1024d) or "voyage-2" (1536d)
• embedding.dimensions - Vector dimensions (optional, auto-detected for known models)
• embedding.apiKey - API key for openai/voyage providers
• embedding.baseUrl - Custom endpoint (optional, defaults to provider's official API)

Storage

• dataDir - Directory for SQLite database storage (default: "~/.local/share/opencode/memory")
• dedupThreshold - Similarity threshold for deduplication (0–1, default: 0.15, clamped to 0.05–0.40)

Logging

• logging.enabled - Enable file logging (default: false)
• logging.file - Log file path (default: "~/.local/share/opencode/memory/logs/memory.log")

When enabled, logs are written to the specified file with timestamps. The log file has a 10MB size limit with automatic rotation.

Development

pnpm build      # Compile TypeScript to dist/
pnpm test       # Run tests
pnpm typecheck  # Type check without emitting