Agent skills for Perplexity

Run GPU workloads on Modal — training, fine-tuning, inference, batch processing.

Deploy GPU-enabled applications to Modal with pre-check workflow, optimization patterns, GPU pricing, WebSocket PTY best practices, and…

Comprehensive Modal.com platform knowledge covering all features, pricing, and best practices

Serverless GPU cloud platform for running ML workloads.

Automatically discover Modal platform skills when working with Modal.

Modal and dialog specialist for web applications.

Choose an ML model or provider when a user asks which checkpoint, router, local model, adapter, tokenizer, or inference backend fits a…

Generate the documentation an engineer-built medical-imaging model must carry — a Model Card (Mitchell et al. 2019), a Datasheet for its…

Expert at writing Rails 8 models with proper validations, associations, scopes, callbacks, and concerns.

Use when building the formal model for a Marketing Science manuscript — turning a marketing phenomenon into an analytical (game-theoretic)…

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining.

Resolve model profile (quality/balanced/budget) at orchestration start and map agents to specific models.

Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT.

Recommends where work runs and which tier and supported effort it needs, using ~/.claude/model-profiles.md.

Guide for maintaining the MassGen model and backend registry.

Use when formulating the model and stating results for an Operations Research (OR) manuscript — defining the…

Pick the executor and effort for a unit of work — which Claude tier (Fable 5 / Opus 5 / Sonnet 5 / Haiku 4.5), what effort level, or…

Refresh a user- or project-owned point-in-time guide for model prompting, benchmark-aware routing, execution lanes, and pricing when the…

When to activate: model serving, inference, FastAPI ML, BentoML, Triton, vLLM, model deployment, online inference, batching

Vet the concrete third-party model a study will be built on — this repository, this revision, this checkpoint — not the architecture family.

Apply when writing stateful logic, or when code branches a lot or repeats a shape assumption across files.