opencode-openmemory


Local-first, privacy-focused persistent memory for OpenCode agents using OpenMemory.

A fork of opencode-supermemory, redesigned to work with OpenMemory - an open-source, self-hosted cognitive memory engine that keeps your data on your machine.

Features

- Local-first: All memories stored on your machine via OpenMemory
- Privacy-focused: No data sent to external services
- Automatic context injection: User profile, project memory, and relevant memories injected into conversations
- Explicit & implicit memory capture: Save memories with "remember this" or let the agent extract knowledge automatically
- Scope separation: User-level (cross-project) vs project-level memories
- Context compaction: Smart summarization when context window fills up

Architecture

``
┌───────────────────────────────────────────────────────────────┐
│ OpenCode (Plugin) │
│ - Injection Policy (format, token budget, priority) │
│ - Memory Capture Policy (explicit "remember", implicit) │
│ - Scope Router (user_id, project_id) │
└───────────────────┬───────────────────────────────────────────┘
│
v
┌───────────────────────────────────────────────────────────────┐
│ OpenMemory Server (REST API) │
│ - Store: raw notes / facts / events / snippets │
│ - Index: embeddings + metadata (scope/recency/type) │
│ - Retrieval: hybrid scoring (similarity + salience + decay) │
│ - Default: http://localhost:8080 │
└───────────────────────────────────────────────────────────────┘
`

Installation

$3

`bash
bunx @eddy.soungmin/opencode-openmemory@latest install
`

Or manually add to ~/.config/opencode/opencode.jsonc:

`jsonc
{
"plugin": ["@happycastle/opencode-openmemory@latest"]
}
`

$3

Option A: Docker (recommended)

`bash
git clone https://github.com/CaviraOSS/OpenMemory.git
cd OpenMemory
cp .env.example .env


Edit .env with your OPENAI_API_KEY (for embeddings)


docker compose up --build -d
`

Option B: Manual setup (for development)

`bash
git clone https://github.com/CaviraOSS/OpenMemory.git
cd OpenMemory/backend
npm install
npm run dev # Starts on :8080 by default
`

For more details, see the OpenMemory documentation.

$3

The plugin will automatically connect to OpenMemory REST API at http://localhost:8080.

Configuration

Create ~/.config/opencode/openmemory.jsonc:

`jsonc
{
// OpenMemory REST API URL
"apiUrl": "http://localhost:8080",

// Search settings
"similarityThreshold": 0.6,
"maxMemories": 5,
"maxProjectMemories": 10,
"maxProfileItems": 5,
"minSalience": 0.3,

// Context injection
"injectProfile": true,

// Scope prefix for organizing memories
"scopePrefix": "opencode",

// Default sector for storing memories
"defaultSector": "semantic"
}
`

Usage

$3

On the first message of each session, the plugin automatically injects:

1. User Profile: Cross-project preferences and patterns
2. Project Knowledge: Project-specific memories from the current directory
3. Relevant Memories: Semantically similar memories to the current query

$3

Use trigger phrases to save memories:

`
"Remember that we use Prettier with single quotes"
"Save this: always run tests before committing"
"Keep in mind that the auth service is in /src/lib/auth"
`

$3

The openmemory tool is available with these modes:

| Mode | Description | Arguments |
|------|-------------|-----------|
| add | Store a new memory | content, type?, scope? |
| search | Search memories | query, scope?, limit? |
| profile | View user profile | query? |
| list | List recent memories | scope?, limit? |
| forget | Remove a memory | memoryId, scope? |
| help | Show usage guide | - |

Scopes:
- user: Cross-project preferences and knowledge
- project: Project-specific knowledge (default)

Memory Types:
- project-config: Tech stack, commands, tooling
- architecture: Codebase structure, components, data flow
- learned-pattern: Conventions specific to this codebase
- error-solution: Known issues and their fixes
- preference: Coding style preferences
- conversation: Session summaries

$3

Run the /openmemory-init command to deeply research your codebase and populate memory:

`
/openmemory-init
`

Context Compaction

When the context window fills up (80% by default), the plugin:

1. Injects project knowledge into the summary prompt
2. Triggers OpenCode's summarization
3. Saves the summary as a memory for future sessions
4. Automatically resumes the conversation

Usage with Oh My OpenCode

If you're using Oh My OpenCode, disable its built-in auto-compact hook to let this plugin handle context compaction:

Add to ~/.config/opencode/oh-my-opencode.json:

`json
{
"disabled_hooks": ["anthropic-context-window-limit-recovery"]
}
`

Development

`bash


Clone

git clone https://github.com/happycastle114/opencode-openmemory.git
cd opencode-openmemory

Install dependencies

bun install

Type check

bun run typecheck

Build

bun run build

Development (watch mode)

bun run dev

Local testing with OpenCode

bun run build && opencode --plugin ./dist/index.js
`

Comparison with opencode-supermemory

| Feature | opencode-supermemory | @eddy.soungmin/opencode-openmemory |
|---------|---------------------|-------------------------------------|
| Backend | Supermemory Cloud | OpenMemory (local) |
| Data Location | Cloud | Your machine |
| Privacy | Requires API key | Fully local |
| Cost | API usage fees | Free (self-hosted) |

License

MIT

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature`)
5. Open a Pull Request

Credits

- Based on opencode-supermemory by Supermemory
- Uses OpenMemory by CaviraOSS
- Developed by @happycastle114

Special Thanks

- Oh My OpenCode - This plugin was developed using Oh My OpenCode's powerful agent orchestration capabilities