Loading...
Loading...
Found 1,262 Skills
Orders scheduler. Reads .noodle/mise.json, writes .noodle/orders-next.json. Schedules work orders based on backlog state, plan phases, session history, and task type schedules.
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for...
A session continuity loop where the frog is disposable but the pad is not.
Use when the user needs prompt design, optimization, few-shot examples, chain-of-thought patterns, structured output, evaluation metrics, or prompt versioning. Triggers: new prompt creation, prompt optimization, few-shot example design, structured output specification, A/B testing prompts, evaluation framework setup.
Token-saving terse mode — no filler, no narration, just results
Use when assessing AI/ML systems for prompt injection, jailbreak vulnerabilities, model inversion risk, data poisoning exposure, or agent tool abuse. Covers MITRE ATLAS technique mapping, injection signature detection, and adversarial robustness scoring.
Build AI agents with in-process agent loops using Anthropic or OpenAI APIs, custom tools, MCP servers, and multi-turn conversations
Recommend and customize Megatron Bridge recipes for a user's model, GPU count, and training goal. Indexes library recipes (pretrain/SFT/PEFT) and performance recipes.
Multi-model deep review of the Ralph bd graph and plan via three parallel opencode processes (claude opus, gemini, gpt). Use for high-stakes runs where cross-model consensus reduces single-model bias.
Step-by-step guide to building AI agents from simple chat loops to autonomous multi-agent systems with tools, memory, and event-driven architecture
RNN+Transformer hybrid with O(n) inference. Linear time, infinite context, no KV cache. Train like GPT (parallel), infer like RNN (sequential). Linux Foundation AI project. Production at Windows, Office, NeMo. RWKV-7 (March 2025). Models up to 14B parameters.
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.