Loading...
Loading...
Found 1,195 Skills
Build AI agents with in-process agent loops using Anthropic or OpenAI APIs, custom tools, MCP servers, and multi-turn conversations
Produce a token-bounded context pack from the Obsidian wiki — a compact, structured slice of the most relevant pages for a topic or recent activity, designed for downstream consumption by another agent or skill. Use when the user says "/wiki-context-pack", "make a context pack", "give me a context slice for X", "pack the wiki for my agent", or "bounded context for Y". Different from wiki-query (which answers a question) — this produces reusable input material for a downstream task.
Debug AutoDeploy accuracy regressions vs a reference score (PyTorch backend or published baseline). Use when an AutoDeploy model's eval score is significantly below the reference and the root cause is unknown.
Interactive config wizard for NeMo Evaluator Launcher (NEL). Use when the user wants to create a new evaluation config from scratch, set up an evaluation from existing configs, or modify a NEL config (deployment, tasks, multi-node, interceptors). ALWAYS triggers on mentions of creating configs, setting up evaluations, configuring models for evaluation, or modifying NEL YAML files. Do NOT use for monitoring, debugging, or analyzing already-running evaluations.
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.
Claude Code skill (trtllm-agent-toolkit): implement or extend TensorRT-LLM AutoDeploy fusion transforms under transform/library/ in a TensorRT-LLM checkout. Prefer existing kernels and custom ops; use Triton only when no viable existing-kernel path exists. Use ad-graph-dump for AD_DUMP_GRAPHS_DIR workflows. Covers TRT-LLM paths, registry, default.yaml registration, graph validation, tests, and a review checklist — without prescribing profiling tools or throughput targets.
Generate a source-backed starting `trtllm-serve --config` YAML for basic aggregate single-node PyTorch serving, aligned with checked-in TensorRT-LLM configs and deployment docs. Preserves explicit latency / balanced / throughput objectives. Excludes disaggregated, multi-node, and non-MTP speculative configs.
Integrate TileGym kernels into Hugging Face `transformers` models by replacing the library's submodule(s) and certain class(es)' implementations, and patching certain class(es)' init/forward/load weight methods prior to instantiating models. Used when the user requires integrating TileGym kernels into `transformers` models.
Creative-mode PPT pipeline. One full-page 16:9 PNG per slide. LLM / VLM calls go through sn-ppt-standard/lib/model_client.py (shared thin client). Text-to-image (the actual png rendering) goes through sn-image-base/scripts/sn_agent_runner.py. Expects task_pack.json + info_pack.json already written by sn-ppt-entry.
ADBPG Knowledge Base Management: Create knowledge bases, upload documents, search, Q&A. Triggers: "knowledge base", "document library", "document upload", "knowledge search", "RAG", "Q&A", "embedding", "ADBPG", "AnalyticDB PostgreSQL"
Iterate on RAG systems with structured evals instead of eyeballing. This skill should be used when the user is tuning a RAG pipeline — changing retrieval prompts, swapping models, adjusting chunking, or debugging poor answers — and wants a cheap, ranked set of experiments with cost tracking and structured feedback on the stack. Also use when the user asks "how do I know if my RAG is working?", "this RAG eval is burning money", or "what should I try next on retrieval?".
Deduplicate and synthesize raw concept stubs into a tiered intellectual map (T1 Canon to T4 Riff), tracing idea evolution across sources over time. Transforms thousands of raw concept pages into a curated intellectual fingerprint.