Total 56,862 skills, AI & Machine Learning has 9448 skills
Showing 12 of 9448 skills
Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested. Integrated into Cavekit: enabled by default for build, inspect, and subagent phases via caveman_mode config. See scripts/bp-config.sh for caveman_mode and caveman_phases.
Sets up or repairs the AGENTS.md source-of-truth pattern for any project. Creates a well-structured AGENTS.md with real stack info auto-detected from the project, then wires all AI config satellites (.claude/CLAUDE.md, .github/copilot-instructions.md, .agents/rules/, MEMORY.md) to point to it. Eliminates duplication. Always runs in plan mode — asks before acting. Use this skill whenever the user mentions AGENTS.md, agent config, source of truth for AI rules, setting up Claude/Copilot/Cursor for a project, fixing duplicate AI instructions, or wants to consolidate AI configuration files. Trigger even if the user just says "set up agents" or "fix my AI config".
Landing AI integration. Manage data, records, and automate workflows. Use when the user wants to interact with Landing AI data.
Use when generating videos from images with DashScope Wan 2.7 image-to-video model (wan2.7-i2v). Use when implementing first-frame video generation, first+last frame interpolation, video continuation, or audio-driven video synthesis via the video-synthesis async API.
Run a structured multi-perspective council on a hard decision, design choice, debugging question, strategy problem, or tradeoff. Use when the user wants multiple viewpoints, explicit cross-examination, and a compact final verdict.
Run existing ShinkaEvolve tasks with the `shinka_run` CLI from a task directory (`evaluate.py` + `initial.<ext>`). Use when an agent needs to launch async evolution runs quickly with required `--results_dir`, generation count, and strict namespaced keyword overrides.
Run vLLM performance benchmark using synthetic random data to measure throughput, TTFT (Time to First Token), TPOT (Time per Output Token), and other key performance metrics. Use when the user wants to quickly test vLLM serving performance without downloading external datasets.
Provides guidance for PyTorch-native agentic RL using torchforge, Meta's library separating infra from algorithms. Use when you want clean RL abstractions, easy algorithm experimentation, or scalable training with Monarch and TorchTitan.
Shared optimization guidance plus cuTile Python DSL-specific overlays. Use when: (1) selecting optimizations for a cuTile Python DSL kernel, (2) checking cuTile-specific implementation traps, (3) deciding whether a profiling finding belongs in shared knowledge or a cuTile overlay, (4) updating cuTile Python DSL optimization docs, (5) reviewing how a shared pattern maps to cuTile.
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing.
Compare Replicate models by cost, speed, quality, and capabilities.
This skill should be used when the user asks to maintain an Obsidian knowledge base for a research project, import an existing research repository into Obsidian, keep project memory or daily notes synchronized, summarize project context into durable notes, or update experiments, results, papers, writing, and plans in an Obsidian vault without requiring MCP.