Skip to main content
Glama
README.md3.57 kB
# PowerMem MCP Server PowerMem MCP Server - Model Context Protocol server for PowerMem memory management. English | [简体中文](powermem_mcp_server_CN.md) ## Prerequisites Before using the PowerMem MCP Server, please ensure: 1. **PowerMem is installed**: The server requires PowerMem to be installed. You can install it via: ```shell pip install powermem ``` 2. **Configuration file exists**: Create a `.env` file in your working directory or project root with PowerMem configuration. The server will automatically search for `.env` files in: - Current working directory's `.env` - Project root directory's `.env` - `examples/configs/.env` You can copy the `.env.example` file as a template: ```shell cp powermem_mcp/.env.example .env ``` Then edit the `.env` file and configure the following key settings: - **Database Provider**: Choose from `sqlite`, `oceanbase`, or `postgres` - **LLM Provider**: Choose from `qwen`, `openai`, or `mock` - **Embedding Provider**: Choose from `qwen`, `openai`, or `mock` - **API Keys**: Set your LLM and Embedding API keys - **Agent Configuration**: Configure memory management settings - **Intelligent Memory**: Enable/configure Ebbinghaus forgetting curve settings - **Performance**: Adjust batch sizes, cache settings, and search parameters - **Security**: Configure encryption and access control - **Logging**: Set logging levels and file paths The `.env.example` file contains detailed comments and example configurations for different use cases. For PowerMem installation and configuration, please refer to: [PowerMem Documentation](https://powermem.ai/docs) ## Startup ### Support for multiple types of MCP You can start PowerMem MCP with different protocols using the following commands: ```shell uvx powermem-mcp sse # sse mode, default port 8000 (recommended) uvx powermem-mcp stdio # stdio mode uvx powermem-mcp sse 8001 # sse mode, specify port 8001 uvx powermem-mcp streamable-http # streamable-http mode, default port 8000 uvx powermem-mcp streamable-http 8001 # streamable-http mode, specify port 8001 ``` ## Usage Use with MCP Client, must use a client that supports Prompts, such as: Claude Desktop. Before entering a request, you need to manually select the required Prompt, then enter the request. Claude Desktop config example: ```json { "mcpServers": { "powermem": { "url": "http://{host}:8000/mcp" } } } ``` ## Available Tools The PowerMem MCP Server provides the following memory management tools: - **add_memory**: Add new memory to storage. Supports string, message dict, or message list format. Can use intelligent mode for automatic inference. - **search_memories**: Search memories by query text with optional filters, limit, and similarity threshold. - **get_memory_by_id**: Get a specific memory by its ID. - **update_memory**: Update the content and metadata of an existing memory. - **delete_memory**: Delete a specific memory by its ID. - **delete_all_memories**: Batch delete memories by user_id, agent_id, or run_id. - **list_memories**: List all memories with pagination support (limit and offset) and optional filters. ## Community When you need help, you can find developers and other community partners at [https://github.com/oceanbase/powermem](https://github.com/oceanbase/powermem). When you discover project defects, please create a new issue on the [issues](https://github.com/oceanbase/powermem/issues) page. ## License For more information, see [LICENSE](LICENSE).

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/oceanbase/mcp-oceanbase'

If you have feedback or need assistance with the MCP directory API, please join our Discord server