Vibetape

VibeTape is an MCP server that records development moments, coordinates multiple agents, and provides intelligent context management and handoffs compatible with any MCP-enabled client. It exposes a rich set of tools for capturing wins, failures, decisions, and notes, as well as task management, actor/agent registration, and context scoring. You can use it to build multi-agent workflows where agents share a persistent, queryable history of actions and outcomes, with smart context selection to stay within token budgets. The system is designed to integrate with LangGraph-compatible payloads and supports RETEX-style prescriptive insights for improving future work.

Prerequisites:

• Node.js and npm installed on the host
• Git available to clone the repository

1. Clone the repository:

git clone https://github.com/sambaleuk/Vibetape-MCP-Server.git
cd Vibetape-MCP-Server

2. Install dependencies:

npm install

3. Build the project (if applicable):

npm run build

4. Run the server locally (example):

# Ensure environment variables are set as needed
OPENAI_API_KEY=your-key-here npm start

5. Verify the server is running and accessible via your MCP client by adding the server entry (see mcp_config) and testing basic MCP commands like mark_moment, register_actor, and create_task.

Tips and common considerations:

• Set OPENAI_API_KEY (and any other required API keys) in your environment or in your MCP client configuration.
• Update the cwd and paths in mcp_config to point to your actual deployment location.
• The server supports LangGraph-compatible handoffs and RETEX generation; ensure your MCP client can send and receive these payloads.
• Watch for token budget limits when selecting context; use context_relevance_score and evaluate_context_window tooling when available.
• If you see connection issues, verify that the server process has started successfully and that the client is configured with the correct command, args, and working directory.