Skill UI

This memory bridge ingests Claude Code's local auto-memory markdown files and structures them into AgentDB. It generates high-fidelity ONNX embeddings, handles deduplication using cosine similarity, and indexes the memories in a scalable namespace. This enables seamless, unified semantic search across all historical sessions and projects, significantly enhancing RAG capabilities.

Uses the Monitor tool to stream real-time events from a swarm system instead of inefficient polling. It provides live observability for key actions like agent spawns, task completions, memory writes, and health status. This is ideal for continuous monitoring and debugging of complex, distributed processes.

This tool manages RVF (Ruflo Vector Format) files, providing comprehensive functions for handling, transferring, and backing up agent memories. It supports listing, exporting, importing, migrating, and generating statistics for complex vector embeddings and associated metadata, ensuring knowledge portability across different projects and platforms.

This toolset allows users to save and restore complete agent sessions, ensuring context continuity across multiple conversations. It captures the agent's entire state, including memory, configuration, task progress, and learning metrics. Use it when pausing work and needing to resume later with full context, or for checkpointing long-running tasks.

Tree Ring Memory is a local-first, framework-agnostic layer for managing durable knowledge in AI agents. It provides structured methods for recalling past decisions, recording evidence, auditing project state, and deliberately forgetting information, ensuring the core project memory remains clean and actionable without relying on raw conversation transcripts.

Ruflo is a powerful, cross-agent orchestration layer designed to manage complex, multi-step AI coding workflows. It supports coordinating large swarms of AI agents, managing persistent memory through semantic search, and executing sophisticated pipelines using 30+ specialized plugins. Ideal for large-scale development tasks, system benchmarking, and deep research requiring coordinated agent effort.

This skill performs a comprehensive, setup-wide health check for agentic developers. It audits various components—including skills, memory, tools, configuration, and workflow habits—to identify inefficiencies and potential improvements. It provides a detailed report and guides the user through applying targeted fixes, ensuring the agent environment remains lean and optimized.

SkillOpt-Sleep enables an AI agent to undergo a simulated "sleep cycle" for continuous, measurable self-improvement. It harvests past sessions, mines recurring tasks, and re-runs them offline. It then consolidates findings into the agent's memory and skills, ensuring only validated improvements are adopted. This process maximizes long-term competence from short-term interactions.

The Unified Memory system provides a durable, inspectable context layer for sharing state and handoffs between multiple autonomous agents (harnesses) like Claude, Codex, and Hermes. It allows agents to save work state, transfer context, resume tasks, and search shared project knowledge across different scopes (project, team, user).

A comprehensive guide for positioning continual learning and lifelong agent research for the Conference on Lifelong Learning Agents (CoLLAs). This tool helps authors strategically re-frame their work, assess its fit, and optimize the narrative to meet the specific evidence bar of top CS venues focusing on nonstationary learning, adaptation, and memory. It covers scope calibration, required evidence types, and comparison with related ML conferences.

This plugin integrates the OpenViking context database, providing powerful long-term memory, session archiving, and semantic recall for OpenClaw agents. It enables agents to query, debug, and manage persistent knowledge bases, significantly enhancing conversational depth and complex tool usage.

This skill allows agents to retrieve and inject relevant operational memories from the OpenViking experience library into the current prompt context. It is designed for complex, multi-step tasks involving tool usage, code operations, or decision-making workflows. By searching and reading past successful executions, the agent improves context awareness and decision accuracy, moving beyond simple Q&A.