Loading...
Loading...
Found 36 Skills
Memory-efficient fine-tuning with 4-bit quantization and LoRA adapters. Use when fine-tuning large models (7B+) on consumer GPUs, when VRAM is limited, or when standard LoRA still exceeds memory. Builds on the lora skill.
Crafting effective prompts for LLMs. Use when designing prompts, improving output quality, structuring complex instructions, or debugging poor model responses.
Implement OpenAI Harness Engineering practices in any repository. Use when setting up or refactoring agent-first workflows, writing or upgrading AGENTS.md and PLANS.md, creating deterministic smoke/test/lint/typecheck harness commands, defining strict architecture boundaries and data-shape contracts, wiring observability from day 1, and adding entropy-control checks plus CI automation for reliable autonomous runs.
Strategies for managing LLM context windows effectively in AI agents. Use when building agents that handle long conversations, multi-step tasks, tool orchestration, or need to maintain coherence across extended interactions.
Building and training neural networks with PyTorch. Use when implementing deep learning models, training loops, data pipelines, model optimization with torch.compile, distributed training, or deploying PyTorch models.
Running and fine-tuning LLMs on Apple Silicon with MLX. Use when working with models locally on Mac, converting Hugging Face models to MLX format, fine-tuning with LoRA/QLoRA on Apple Silicon, or serving models via HTTP API.
Project planning and management with CodeSpring. Use when the user wants to work with CodeSpring projects, tasks, PRDs, mindmaps, or analyze a codebase for project planning. Handles workspace selection, project linking, task management, and syncing findings to CodeSpring.
Creates dbt models following project conventions. Use when working with dbt models for: (1) Creating new models (any layer - discovers project's naming conventions first) (2) Task mentions "create", "build", "add", "write", "new", or "implement" with model, table, or SQL (3) Modifying existing model logic, columns, joins, or transformations (4) Implementing a model from schema.yml specs or expected output requirements Discovers project conventions before writing. Runs dbt build (not just compile) to verify.
Safely refactors dbt models with downstream impact analysis. Use when restructuring dbt models for: (1) Task mentions "refactor", "restructure", "extract", "split", "break into", or "reorganize" (2) Extracting CTEs to intermediate models or creating macros (3) Modifying model logic that has downstream consumers (4) Renaming columns, changing types, or reorganizing model dependencies Analyzes all downstream dependencies BEFORE making changes.
Optimizes Snowflake SQL query performance from provided query text. Use when optimizing Snowflake SQL for: (1) User provides or pastes a SQL query and asks to optimize, tune, or improve it (2) Task mentions "slow query", "make faster", "improve performance", "optimize SQL", or "query tuning" (3) Reviewing SQL for performance anti-patterns (function on filter column, implicit joins, etc.) (4) User asks why a query is slow or how to speed it up
Understanding Reinforcement Learning from Human Feedback (RLHF) for aligning language models. Use when learning about preference data, reward modeling, policy optimization, or direct alignment algorithms like DPO.
Patterns and architectures for building AI agents and workflows with LLMs. Use when designing systems that involve tool use, multi-step reasoning, autonomous decision-making, or orchestration of LLM-driven tasks.