Skill UI

This skill loads and manages company context for C-suite advisors. It reads context files, detects stale data older than 90 days, enriches context during conversations, and enforces privacy rules before external API calls. Use it to start advisor sessions, refresh context, or ensure data privacy.

Use this command to set up a structured multi-agent collaboration environment within AgentHub. It allows you to define the core task, specify the required number of agents, and configure detailed evaluation criteria (metrics, directions, and commands) for both quantitative benchmarking and qualitative LLM judging. This is the starting point for any agent competition or complex AI workflow.

Provides guidance for organizing and navigating Jupyter Notebook workflows, focusing on structuring analysis, experiments, and reports within interactive Python sessions.

This tool allows users to explicitly save critical, project-specific facts, conventions, or debugging insights to a dedicated auto-memory file. Use it when the knowledge is too important or specialized to rely on the AI's automatic context capture. It ensures persistent recall of architectural decisions, tool gotchas, or team preferences across multiple sessions and projects.

This skill provides best practices for managing application state and ensuring data persistence within large-scale applications, particularly for systems using Makepad architecture (Robrix/Moly). It details core structures like AppState, techniques for passing state through widget trees (Scope), and handling persistent data serialization (serde) across sessions, making it essential for building robust, multi-session UIs.

The /hub:status command provides a comprehensive overview of an AgentHub session. It displays the current state of agent branches, commit counts, frontier status, and recent board updates. This is useful for monitoring ongoing collaborative AI tasks, allowing users to track progress and identify which agents are active or have completed their tasks.

This agent ensures seamless continuity across multiple work sessions. It automatically captures, compresses, and stores critical project information—including summarized discussions, architectural decisions, pending tasks, modified files, and technical discoveries. When a new session starts, it loads a comprehensive briefing, allowing users to quickly resume work without losing track of the project's current state.

Context Guardian preserves critical context before Claude Code’s automatic compression, collecting verified snapshots, detecting integrity issues, and briefing transitions so heavy-technical sessions keep every essential decision and dependency intact.

This utility converts a single, high-level objective into a comprehensive, step-by-step build plan suitable for complex, multi-agent, or multi-session engineering projects. It automatically handles dependency mapping, detects parallel steps, and structures each step with self-contained context briefs, ensuring any new agent can execute it without prior knowledge. Ideal for large-scale refactoring or feature development spanning multiple PRs.

A forensic tool for performing post-mortem analysis on AI-assisted coding sessions. It goes beyond simply stating what was done, determining *why* work deviated (e.g., spec ambiguity, repo friction, agent error), identifying scope creep, and classifying rework patterns. It provides evidence-backed recommendations to improve prompts, codebase health, and future development workflows.

RecallMax dramatically enhances AI agent memory capabilities by cleanly injecting massive external contexts (500K-1M tokens) without hallucination drift. It automatically summarizes lengthy conversations, preserving tone, intent, and key facts. Furthermore, it compresses multi-turn interaction histories into high-density token sequences, ensuring agents maintain perfect recall across complex and long-running sessions.

This skill simulates Pass-the-Ticket (PtT) attacks, a critical post-exploitation technique. It involves dumping active Kerberos tickets (TGT/TGS) from memory (LSASS) on a compromised host and injecting them into another session. This allows an attacker to impersonate the ticket owner and perform lateral movement or escalate privileges without needing the user's password. Use exclusively for authorized red-teaming and security auditing.