Latest Agent Skills

Audit and fix HTML metadata including page titles, meta descriptions, canonical URLs, Open Graph tags, Twitter cards, favicons, JSON-LD structured data, and robots directives.

Routes UI-related tasks to the smallest useful UI skill context via the ui-skills CLI.

Enforces an opinionated UI baseline to prevent AI-generated interface slop, covering Tailwind CSS, accessible components, interaction patterns, animation constraints, typography, layout, and performance.

Guides configuration of n8n nodes with operation-aware field requirements, property dependencies, and progressive discovery patterns.

Provides proven architectural patterns for building n8n workflows, covering webhook processing, HTTP API integration, database operations, AI agent workflows, batch processing, and scheduled tasks.

Interpret n8n validation errors and warnings, guide iterative fixes, and explain validation profiles, error types, and auto-sanitization.

Validate n8n expression syntax and fix common errors when writing {{}} expressions, accessing $json/$node variables, and mapping data between nodes.

Routes n8n workflow-building tasks to the correct skill, with defaults and red-flag guidance for reliable n8n automation.

Provides expert guidance for using n8n-mcp MCP server tools to build, validate, and manage n8n workflows efficiently.

Write JavaScript code in n8n Code nodes for data transformation, API calls, and workflow logic.

Handle files and binary data in n8n workflows correctly, covering the $binary vs $json split, reading/writing binary, preserving binary across transforms, and the agent-tool binary boundary.

Manage multiple n8n instances over MCP by discovering, switching, and verifying the target instance before reads and writes.

Makes n8n workflow failures loud, structured, and recoverable by wiring per-node error outputs, retry logic, error-trigger workflows, and proper HTTP response shapes.

Build reusable, composable n8n sub-workflows with typed inputs, stateless or stateful design, and discoverable naming conventions.

Deploy a production self-hosted n8n instance on a fresh Linux VM over SSH, using Docker Compose behind a Caddy reverse proxy with automatic HTTPS.

Write Python code in n8n Code nodes, covering data access patterns, return formats, and standard library limitations.

Write and debug code inside n8n's Custom Code Tool node, with guidance on input modes, return formats, error handling, and common pitfalls.

Design n8n AI agents with best practices for node selection, sub-node wiring, tool design, structured output, and memory management.

Capture and organize general facts about the user by topic, such as preferences, background, and goals, to personalize interactions.

Records daily activities, progress, decisions, and learnings in structured chronological markdown files.

Install and configure Acontext, a memory layer for AI agents that provides persistent sessions, file storage, and skill management.

Generate brand-quality product images by selecting a photography mode and answering a short interview; the backend enhances the prompt and submits it to GPT Image 2.

Generate marketplace product image cards, including main images, secondary product shots, and A+ content modules, via the Higgsfield CLI.

Build, edit, and deploy full-stack websites or web apps using the Higgsfield CLI, with two distinct product types: standalone websites or Higgsfield-integrated apps.

Generate images, videos, 3D assets, and audio via the Higgsfield AI CLI, including Marketing Studio ads and Virality Predictor analysis.

Train a personalized identity model on a person's face for use in Higgsfield's image and video generation.

Applies cognitive science frameworks for creative thinking to CS and AI research ideation, using combinatorial creativity, analogical reasoning, constraint manipulation, and other empirically grounded strategies.

Records research provenance as a post-task epilogue, scanning conversation history to extract decisions, experiments, dead ends, claims, heuristics, and pivots, and writing them into the ara/ directory with provenance tags.

Guides researchers through structured ideation frameworks to discover high-impact research directions.

Performs a semantic epistemic review of Agent-Native Research Artifacts, scoring six dimensions and producing a constructive report with a recommendation.

Inspect and manipulate transformer internals via HookPoints and activation caching for mechanistic interpretability research.

Generates conference presentation slides (Beamer LaTeX PDF and editable PPTX) from a compiled paper with speaker notes and talk script.

Compress large language models using knowledge distillation from teacher to student models, covering temperature scaling, soft targets, reverse KLD, logit distillation, and MiniLLM training strategies.

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques for 1.5-3.6× speedup without quality loss.

Compiles research inputs—PDFs, code, logs, notes—into structured Agent-Native Research Artifacts with cognitive and physical layers.

Provides paragraph-level structural blueprints for writing systems papers targeting OSDI, SOSP, ASPLOS, NSDI, and EuroSys, including page allocation, writing patterns, and venue-specific checklists.

Organizes PyTorch code with a Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks, and minimal boilerplate. Scales from laptop to supercomputer with the same code.

Evaluates code generation models across HumanEval, MBPP, MultiPL-E, and 15+ benchmarks with pass@k metrics. Use when benchmarking code models, comparing coding abilities, testing multi-language support, or measuring code generation quality.

Train and analyze Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features for mechanistic interpretability research.

Run interpretability experiments on neural network internals using nnsight, with optional NDIF remote execution for massive models.

Perform causal interventions on PyTorch models using pyvene's declarative framework for causal tracing, activation patching, and interchange intervention training.

Fast tokenization for NLP using Rust-based tokenizers supporting BPE, WordPiece, and Unigram algorithms, with training, alignment tracking, and padding/truncation.

Generates publication-quality figures for ML papers, including architecture diagrams via Gemini and data-driven charts via matplotlib/seaborn.

Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies for maximum GPU efficiency.

Train AI models to be harmless through self-critique and AI feedback using a set of constitutional principles, without requiring human labels for harmful outputs.

Draft publication-ready ML/AI papers for top conferences (NeurIPS, ICML, ICLR, ACL, AAAI, COLM) by exploring research repositories, structuring arguments, verifying citations via APIs, and formatting submissions.

Compress large language models by 40-60% with minimal accuracy loss using one-shot pruning techniques like Wanda and SparseGPT, enabling faster inference and deployment on constrained hardware.

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining, covering SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace, covering architectures, routing, load balancing, and expert parallelism.

Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques for processing long documents and implementing efficient positional encodings.

Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag) using standardized prompts and metrics. Supports HuggingFace, vLLM, and API backends.

Provides battle-tested PyTorch training recipes for LLMs, vision, diffusion, and biomedical domains, covering training loops, optimizer selection, LR scheduling, mixed precision, and debugging.

Add programmable safety guardrails to LLM applications at runtime, including jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, and toxicity detection.

Add distributed training support to any PyTorch script with minimal code changes using a unified API for DDP, DeepSpeed, FSDP, and mixed precision.

Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency on NVIDIA GPUs (A100/H100).

Scales machine learning training from single GPU to multi-node clusters with minimal code changes. Supports PyTorch, TensorFlow, and HuggingFace with built-in hyperparameter tuning, fault tolerance, and elastic scaling.

Provides expert guidance for distributed training with DeepSpeed, covering ZeRO optimization stages, pipeline parallelism, FP16/BF16/FP8, 1-bit Adam, and sparse attention.

Expert guidance for implementing GRPO/RL fine-tuning with TRL for reasoning and task-specific model training.

Extract structured data from LLM responses with Pydantic validation, automatic retries, and streaming support across multiple providers.