715 skills · sorted by relevance
Develop lightweight CUDA operations for SGLang using JIT, which compiles code on first use. A tensor-scaling example explains implementation conventions, input validation, memory access helpers, and when to choose this approach.
Configure safety checks for LLM applications with NeMo Guardrails. Covers attempts to bypass safety rules, input and output validation, masking personal information, and checking factual claims through programmable rules and integrations.
Build safety checks for LLM prompts and responses using Meta's LlamaGuard moderation model. Classify potentially harmful content, block flagged messages, serve checks through an API, and integrate them with NeMo Guardrails.
Guides large language model reinforcement learning with miles, covering environment setup, low-precision training, consistency between training and inference, and speculative decoding, where a smaller model proposes tokens for a larger model to verify.
Write, review, and upgrade Effect v4 code for Composio CLI and its related packages. It guides service definitions, typed errors, command APIs, and version alignment, checking unfamiliar APIs against source files and installed type definitions.
Manage complex software and SaaS projects by assessing schedules, budgets, quality, risks, and team capacity. Compare priorities across projects, identify resource constraints, and turn portfolio analysis into executive reports with recommendations for decisions and follow-up.
Build quarterly product plans by translating company goals into product and team OKRs, meaning objectives with measurable results. Generate linked goal hierarchies, check alignment across teams, and review how responsibilities are distributed.
Read, create, and review PDFs with checks of both content and appearance. Uses Python tools for document generation and text extraction, then inspects rendered page images for spacing, alignment, and readability.
Write, revise, and review beads user documentation using consistent terminology, explanations, diagrams, and navigation. Follow project rules for generated pages and release alignment, with checks for links, freshness, and documentation consistency.
Compare feature ideas by expected impact, required effort, uncertainty, and fit with product goals. Get five ranked recommendations, with reasons for each choice, trade-offs, and explanations of which ideas should receive lower priority.
Organize customer feature requests into themes and rank the top three against product goals, customer value, effort, and risk. For each recommendation, explain the underlying need, possible alternatives, uncertain assumptions, and simple ways to test them.
Review game interface specifications, HUD designs for information shown during play, and shared interaction rules before implementation. Check completeness, accessibility, and consistency with game design documents, then receive a verdict with specific issues and missing review inputs.
Maintain existing tests for Walkontable, Handsontable's table rendering engine. The skill explains its separate test commands and covers frozen rows and columns, visible area calculations, scroll alignment, and pitfalls that can hide failures or produce misleading results.
Reviews Google Cloud spending and resource usage to identify waste, recommend suitable resource sizes, compare discount options, and establish ongoing cost management using the cost optimization guidance in the Google Cloud Well-Architected Framework.
Compare a page with search results to identify mismatches in page type, content, and user needs. Receive alignment scores, assessments for different audience profiles, prioritized improvements, and optional outlines of current and proposed page layouts.
Turn a paper outline and research narrative into a section-by-section LaTeX draft. Incorporate figures and verified bibliography entries, adapt formatting to the target venue, and review clarity, evidence alignment, and consistency across sections.
Add safety checks to LLM applications with NeMo Guardrails. This skill guides input and output validation, detection of attempts to bypass safety rules, personal information masking, and checks for unsupported or inaccurate answers.
Guide large language model training with OpenRLHF using human preference feedback. Covers training method selection, distributed GPU setup, reward model training, and memory tuning, with examples for PPO, GRPO, and DPO.
Use Meta's Prompt Guard to detect malicious instructions in user prompts and external data. This skill guides input filtering, batch checks for retrieved documents, detection threshold adjustments, and splitting long text for inspection.
Guides large language model reinforcement learning with miles, including MoE models that route work to specialized components. Covers low-precision training, matching training and inference behavior, and accelerating training response generation with draft models.
Build safety checks for LLM prompts and responses using Meta's LlamaGuard. This skill covers content classification, batch moderation, serving a moderation API, and connecting checks to NeMo Guardrails.
Prepare five-scene prompts and parameter tables for product videos featuring a digital presenter in InfiniteTalk. Specify timed movements, check character consistency, and plan audio and image preparation for a 25-second vertical video.
Process Excel workbooks by saving Python scripts to files and running them through the shell. Covers multiple sheets, data transformations, formulas, formatting, and error reporting for complex spreadsheet tasks.
Check deliverables before marking a task complete. Review the requested file format, confirm every reference file was used, and compare the result with the original request to catch omissions and work that misses the goal.