Integrates Mem0 universal memory layer with OpenCode for intelligent long-term memory capabilities.

Traditional AI assistants have no memory - they forget everything after each conversation. Mem0 changes this by providing:

• Multi-Level Memory: Adaptive personalization across users, sessions, and agents
• Semantic Search: Retrieve relevant memories using natural language queries
• Cost Efficient: Reduces token usage by 90% compared to full context
• Developer-Friendly: Intuitive API, cross-platform SDKs, and fully managed service option
• Continuous Learning: Remembers user preferences and adapts to individual needs over time

• Add Memory: Store user preferences, facts, and information
• Semantic Search: Retrieve relevant memories using natural language
• Get All Memories: Retrieve all memories for a user, agent, or session
• Delete Memory: Delete specific memories or all matching memories
• Memory History: Get all versions and changes of memories
• Session Compaction: Auto-inject memory context during session compaction

# Clone the repository
git clone https://github.com/allbugterminator/opencode-mem0-plugin.git

# Copy to global plugin directory
mkdir -p ~/.config/opencode/plugins/
cp -r opencode-mem0-plugin ~/.config/opencode/plugins/mem0

# Or place in project-level
mkdir -p .opencode/plugins/
cp -r opencode-mem0-plugin .opencode/plugins/mem0

Sign up at Mem0 Platform to get your API key.

Create mem0-config.json in the plugin directory:

{
  "apiKey": "your-mem0-api-key"
}

Or via environment variables:

export MEM0_API_KEY=your-api-key

cd ~/.config/opencode/plugins/mem0
bun install 
# or use: npm install

The plugin will load automatically on next startup.

Store information in Mem0 memory. Use to remember facts, preferences, skills, and important context.

mem0_add memory="User prefers using VS Code for development" userId="user123"
mem0_add memory="User prefers dark theme" userId="user123" metadata={"category": "preference"}

Retrieve relevant memories using natural language query. Fast embedding-based retrieval.

mem0_search query="What development tools does user prefer" userId="user123" limit=5
mem0_search query="User's work preferences" agentId="assistant" sessionId="session1"

Get all memories for a user, agent, or session. Complete listing of stored memories.

mem0_get_all userId="user123" limit=10
mem0_get_all sessionId="session1"

Delete a specific single memory.

mem0_delete memoryId="mem_xxx"

Delete all memories matching criteria. Must specify at least one filter.

mem0_delete_all userId="user123"
mem0_delete_all sessionId="session1"

Get history of a memory - all changes and versions.

mem0_get_history memoryId="mem_xxx"

User: I prefer receiving project progress summaries in the afternoon.
Agent: I'll remember that you prefer afternoon summaries. Let me store this preference.

Call mem0_add to store this preference for future reference.

When user asks "What did I work on last time?", the Agent:

1. Calls mem0_search to query relevant memories
2. Gets memories about past projects
3. Provides personalized, context-aware response

During session compaction, the Mem0 plugin automatically retrieves relevant memories and injects them into context, ensuring:

• New agents can resume from where they left off
• Important context is not lost
• Conversation continuity is maintained

┌─────────────────┐     ┌─────────────────┐
│    OpenCode     │     │     Mem0        │
│      Agent      │────►│    Memory       │
│                 │     │     Service     │
└─────────────────┘     └─────────────────┘
        │                       │
        │ Tools:                │ Storage:
        │ - mem0_add            │ - Vector Store
        │ - mem0_search         │   (Qdrant, Chroma, etc.)
        │ - mem0_get_all        │
        │ - mem0_delete         │ LLM Providers:
        │                       │ - OpenAI
                               │ - Anthropic
                               │ - Custom

To use self-hosted Mem0 service:

{
  "apiKey": "your-api-key",
  "host": "http://localhost:8000"
}

See Mem0 docs for detailed self-hosted guide.

• Bun (to run OpenCode)
• Mem0 npm package: mem0ai
• Mem0 API Key (for cloud) or self-hosted Mem0 server

Check if plugin directory exists:

ls ~/.config/opencode/plugins/

Ensure mem0-config.json is correct, or set environment variables:

echo $MEM0_API_KEY

Manually install dependencies:

cd ~/.config/opencode/plugins/mem0
bun add mem0ai

• Verify userId, agentId, sessionId parameters are correct
• Check if Mem0 Cloud account has sufficient quota

• Mem0 - Universal memory layer for AI Agents
• OpenCode Plugins Docs - OpenCode plugin development guide
• mem0ai npm - Mem0 JavaScript SDK

Apache License 2.0 - See LICENSE for details.

If you find this plugin useful, please star the Mem0 repository!