MCP App StoreMCP App Store
from claude-skill-registry431
Train and fine-tune language models on Hugging Face Jobs using TRL (SFT, DPO, GRPO) with Trackio monitoring and automated Hub push. Includes scripts, cost estim
Provides step-by-step guidance and templates to run TRL (Transformer Reinforcement Learning) training workflows on Hugging Face Jobs. Covers supervised fine-tuning (SFT), Direct Preference Optimization (DPO), GRPO, reward modeling, and converting trained models to GGUF for local deployment. Includes example scripts, PEP 723 inline dependency usage for hf_jobs, Trackio monitoring instructions, and required Hub push configuration so training artifacts are preserved.
Use this skill when users want to fine-tune or RL-train language models on cloud GPUs without local infrastructure, need help selecting hardware and timeouts, want to validate datasets before GPU runs, or need automated conversion to GGUF for local inference. Ideal for planned training jobs, cost estimates, and producing production-ready training scripts.
Primarily for agents that can submit cloud jobs or generate training code (Claude Code/Claude-in-code style assistants). Also useful for developer CLIs that interact with Hugging Face Jobs and CI systems.
TRL Training skill for Hugging Face Jobs — comprehensive guide covering SFT, DPO, GRPO training with UV scripts, cost estimation, dataset validation, and GGUF conversion. No bundled scripts were present in the fetch. SKILL.md is well-structured with clear troubleshooting, hardware guidance, and progressive disclosure via references/. Minor concern: references external script URLs which could drift, but standard for HF ecosystem.
Well-crafted skill for a popular use case. No security issues found. Scripts directory referenced but empty in fetched data — example scripts live in the repo but weren't included in the fetch payload.
Uloop: Execute Dynamic Code
Run small C# snippets in the Unity Editor via the uloop CLI for editor automation tasks like prefab wiring, AddComponent flows, and scene edits—without writing
Bookmarklet Creation
Generates browser-executable JavaScript bookmarklets with strict formatting (IIFE wrapper, block comments) and provides ready-to-install links or installer inst
Overnight — Autonomous Long-Running Coding
Orchestrates long-running coding goals: decomposes objectives into atomic tasks, dispatches isolated worktree workers, verifies acceptance criteria, and merges
Bexio API (Swiss CRM & Invoicing)
Integrate and manage Bexio contacts, quotes, invoices, orders and products via the Bexio API. Useful for CRM and Swiss business document workflows.
Content Research Writer
A writing-partner skill that helps research, outline, draft, cite, and iteratively improve articles, tutorials, and thought pieces.
Agent Hierarchy Diagram
Generate visual hierarchy diagrams (ASCII, Mermaid, GraphML) that show agent roles, levels, and delegation for documentation and onboarding.
Review Pull Request
Automated, structured PR reviewer: gathers metadata, diffs, CI results, dependency changes and provides a concise verdict with testing and documentation recomme
Agent Ops — Testing Workflow
Guidance for designing, running, and analyzing test suites for agents: test isolation, execution patterns, and coverage-based enforcement.
libagent
Agent orchestration library for conversational AI — coordinates LLM completions, memory, tool execution, and multi-turn flows; useful for building chat agents a
Raindrop.io API
Manage Raindrop.io bookmarks, collections, tags and highlights via the Raindrop REST API with helper scripts and examples.
Microsoft Foundry Classic — Expert Guidance
Comprehensive guidance for building, configuring, troubleshooting, and deploying Microsoft Foundry Classic agents and integrations.
Synalinks Framework
Keras-inspired framework for building structured, neuro-symbolic LLM programs with DataModel schemas, modular Programs, and training/optimization tools.
Runtime Communication (research_mvp)
Rules and workflows for messaging, delegation, and task coordination in the research_mvp local multi-agent runtime (leader, researcher, trainer).
Deliberate Practice
Guided framework for accelerating skill acquisition using focused practice, immediate feedback, and progressive challenge—useful for learning technical skills,
libagent
Agent orchestration library for conversational AI — coordinates LLM completions, memory, tool execution, and multi-turn flows; useful for building chat agents a
Relax: Development & Remote Training Debugging
Tools and procedures to develop the Relax project and validate changes by submitting and monitoring remote Ray training jobs (non-blocking, debug-friendly).
Hugging Face Evaluation Manager
Extract, import, and add structured model evaluation results to Hugging Face model cards; run or import benchmark evaluations and generate model-index YAML for
KnowBe4 (Membrane)
Integrate with KnowBe4 via the Membrane CLI to manage users, phishing and training campaigns, groups, reports, and account settings.