muninn

What This Skill Does

Muninn is a sophisticated memory and context-management layer designed for OpenClaw agents, utilizing the Universal Context Protocol (CXP) version 2.3.7. By providing a structured, local-first brain for your AI projects, Muninn allows agents to store, retrieve, and index project-specific knowledge without ever offloading data to external servers. It effectively solves the 'memory amnesia' problem common in long-term AI workflows by creating a persistent index of architectural decisions, file structures, and user preferences within the .muninn directory. The system uses a high-performance Rust-based engine to ensure that context injection is pfeilschnell (lightning-fast) and token-efficient.

Installation

To integrate Muninn into your workspace, execute the following command in your terminal: clawhub install openclaw/skills/skills/endgegnerbert-tech/muninn Once installed, ensure your project directory is prepared to host the .muninn metadata folder. The skill initializes seamlessly as an MCP server, allowing your agent to interface with local project files immediately after the initial setup.

Use Cases

Muninn is ideal for complex software development projects where context is too vast to fit into a single prompt. Use it when:

• Managing large codebases where you need to reference specific architectural patterns established months ago.
• Maintaining consistency in coding style and library choices across multiple agent sessions.
• Creating a persistent 'project knowledge base' that remains accessible even after the agent restarts.
• Reducing token consumption by surgically searching for relevant code snippets rather than reading entire file trees.

Example Prompts

1. "Muninn, run brain_check to initialize our current progress on the authentication module and tell me what the last decided security standard was."
2. "Search_context for all functions related to 'database connection retry' in our codebase so we can apply the new timeout settings."
3. "Add_memory: We have decided to migrate from REST to GraphQL for the user dashboard API to improve frontend data fetching efficiency."

Tips & Limitations

• Pro-Tip: Always execute brain_check at the start of your session. This mandatory orientation ensures the agent is aware of the current state and avoids redundant indexing operations.
• Limitations: Muninn is a strictly local tool. While this provides maximum privacy, it means memories do not sync across different physical machines automatically. Ensure you commit your .muninn folder to your version control system if you want to share project context across different development environments.