Shrimp Task Manager is an AI agent programming framework designed to streamline development workflows by converting natural language into structured, dependency-tracked tasks. It targets AI agents and developers seeking to automate complex programming processes, offering benefits like improved task memory, reduced redundancy, and consistent project standards.

How It Works

Shrimp operates as a sophisticated prompt template system, guiding AI agents through a structured programming lifecycle. It leverages the Model Context Protocol (MCP) to interact with compatible clients like Cursor IDE. Key features include intelligent task decomposition, dependency management, and iterative refinement, enabling agents to perform systematic programming, research, and code generation with enhanced memory and context awareness.

Quick Start & Requirements

• Installation: npx -y @smithery/cli install @cjo4m06/mcp-shrimp-task-manager --client claude or manual npm install && npm run build.
• Prerequisites: Node.js, TypeScript.
• Configuration: Requires setting DATA_DIR (absolute path) and optionally TEMPLATES_USE (e.g., 'en', 'zh') and ENABLE_GUI via environment variables or client configuration.
• Documentation: Prompt Customization Guide, Changelog.

Highlighted Details

• Supports a "Research Mode" for systematic technical investigation and knowledge gathering.
• Features a "Task Memory Function" for long-term storage and retrieval of task history to avoid duplicate work.
• Includes "Project Rules Initialization" to standardize development practices and maintain consistency.
• Offers an optional Web GUI for task management.

Maintenance & Community

The project is maintained by cjo4m06. Further community and roadmap details are not explicitly provided in the README.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Designed to work with MCP-compatible clients, such as Cursor IDE.

Limitations & Caveats

Due to LLM token limits, context may be lost in lengthy conversations, requiring users to restart sessions. The project is optimized for Claude 3.7 and Gemini 2.5, with potential for varying results on other models. The README emphasizes the critical need for absolute paths for DATA_DIR to prevent data loss and system failures.