Loading...
Loading...
Found 38 Skills
Expert guide for deploying, configuring, and optimizing Hermes AI agents with multi-platform support, MCP integration, and production best practices
Optimizes agent context setup. Use when starting a new session, when agent output quality degrades, when switching between tasks, or when you need to configure rules files and context for a project.
Autonomously optimizes end-to-end LLM inference throughput on AMD Instinct GPUs and reports a validated gain, using the Hyperloom multi-agent optimizer. Given a model, framework, workload (TP/EP, concurrency, ISL/OSL, precision), an objective and a time budget, it explores per-workload which levers to pull (serving/config parameters and env, framework enablement and source patches, and hot GPU-kernel rewrites), benchmarks each candidate, and returns the optimization stack that produced the gain. Use when the user wants to make a model serve faster, raise tokens/sec or throughput, optimize or tune vLLM or SGLang on MI300X/MI325X/MI355X, run Hyperloom, run the kernel-agent, quantize-then-optimize with Quark, set up Hyperloom from scratch, or resume a Hyperloom session. Do not use to stand up a server for plain serving, diagnose a broken ROCm install, or run a one-off kernel/benchmark or trace analysis without the optimization loop.
Skill Evolver (Taotie) — Strengthen the target skill by "devouring" and analyzing the advantages of other skills. This skill must be triggered when users intend to: integrate two skills, optimize one skill with another, compare and analyze the pros and cons of two skills, extract the strengths of one skill into another, or express intentions like "feed X to Y", "use X to optimize Y", "integrate these two skills", "devour this skill", "skill evolution", "skill upgrade", "merge skills", etc. Even if users don't explicitly mention "Taotie", this skill should be used as long as it involves capability transfer, comparative analysis, or advantage extraction between two skills.
Build autonomous game-playing agents using AI and reinforcement learning. Covers game environments, agent decision-making, strategy development, and performance optimization. Use when creating game-playing bots, testing game AI, strategic decision-making systems, or game theory applications.
This skill should be used when the user asks to "create a ReAct agent", "build an agent with tools", "implement tool-calling agent", "use dspy.ReAct", mentions "agent with tools", "reasoning and acting", "multi-step agent", "agent optimization with GEPA", or needs to build production agents that use tools to solve complex tasks.
Analyze agent-user interaction transcripts to identify context network maintenance needs and guidance improvements. Use after significant agent interactions or to improve context networks.
Summarize lessons learned from ccbox session logs (projects/sessions/history/skills) so the agent can do better next time. Produce copy-ready instruction updates (project + global) backed by evidence, with optional skill-span context to attribute failures to specific skills. Use when asked to run /ccbox:insights, generate a "lessons learned" memo, or propose standing instructions from session history.
Improve an existing prompt or skill with targeted, minimal-diff edits that preserve its core intent, and return the revised artifact plus a short changelog and tradeoffs note. Use this whenever the user wants to refine, sharpen, tighten, or upgrade an existing prompt or skill, asks to "make it better," or wants a small high-leverage edit instead of a full rewrite — even if they don't explicitly mention tuning.
Create, optimize, update, and validate AGENTS.md files with maximum token efficiency. Use when the user asks to (1) create new AGENTS.md files for any repository, (2) optimize/condense existing AGENTS.md to reduce token count, (3) update/refresh AGENTS.md to sync with codebase changes, (4) validate AGENTS.md quality and completeness, or (5) improve AGENTS.md files to be more effective for AI agents. Always generates token-efficient, condensed output focused on actionable commands and patterns while maintaining model-agnostic language.
After an agentic task completes, perform a retrospective analysis across 6 dimensions (goal alignment, efficiency, decision quality, error handling, communication, reusability). Score performance, identify inefficiency patterns, evaluate skill usage, and produce actionable improvement recommendations. Triggers on "how did it go", "retrospective", "review performance", "what could be better", or after any long agentic task completes.
Analyzes Claude Code session transcripts to evaluate skill portfolio health — routing errors, attention competition between descriptions, and coverage gaps. Generates an interactive HTML report with per-skill health cards, competition matrix, attention budget analysis, and actionable patches. Unlike skill-creator which optimizes individual skills in isolation, skill-auditor optimizes the portfolio as a system, detecting cross-skill attention theft and cascade risks. Use when user says "audit my skills", "skill audit", "run skill-auditor", "analyze skill routing", "check skill competition", "portfolio health", "スキル監査", "スキルの精度を分析", "スキルルーティング分析".