CUDA crashes often hide the inputs and state that caused the failure. This skill uses FlashInfer logging and NVIDIA tools to expose actionable diagnostic evidence.

GPU kernel comparisons are unreliable when timing methods and parameters vary. This skill guides reproducible FlashInfer benchmarks using CUPTI or CUDA events.

Adding a FlashInfer kernel requires coordinated CUDA, TVM-FFI, Python, tracing, testing, AOT, and benchmark changes. This skill provides a complete implementation workflow with repository-specific examples.

Current web research often needs more than short search snippets. This skill runs Firecrawl searches, saves structured results, and can extract full page content for review.

Web pages can be hard to extract cleanly when scripts render important content. This skill uses Firecrawl to scrape URLs into LLM-ready markdown and optional formats.

Local files often need clean text before review, search, or summarization. This skill uses Firecrawl to convert documents into saved markdown for Claude, Codex, and Claude Code.

Teams need reliable alerts when pages, site content, or web search results change. This skill guides Claude, Codex, and Claude Code through Firecrawl monitors with schedules, goals, and notification channels.

Large websites can make the right page hard to find. This skill maps URLs with Firecrawl and filters results by search terms.

Dynamic websites can block simple scraping when content needs clicks, forms, or sessions. This skill uses Firecrawl browser interaction to navigate pages and extract data.

Saving documentation by hand is slow and easy to miss. This skill uses Firecrawl to map pages, download selected content, and organize local files.

Teams need reliable content from many pages without manual copying. This skill guides Firecrawl crawls with scoped paths, limits, progress, and saved results.

Teams need reliable source discovery before scraping or answering questions. This skill guides Firecrawl search integration, result selection, and follow-up extraction decisions.

Apps often need clean content from one known URL. This skill guides a focused Firecrawl /scrape integration for page extraction.

Teams need a clear path from Firecrawl signup to a working application integration. This skill guides credential setup, SDK choice, and smoke testing for Claude, Codex, and Claude Code workflows.

Dynamic pages often hide data behind clicks, forms, and login flows. This skill helps teams choose and scope Firecrawl /interact integrations.

Complex websites often require multi-page navigation before data can be collected. This skill guides Claude, Codex, or Claude Code to run Firecrawl Agent and return structured website data.

Adding Swift Packages to Xcode projects can be error-prone and easy to break. This skill uses a bundled Swift tool to add packages, link products, and handle Firebase plist setup.

Firestore rules often miss update bypasses, ownership checks, and validation gaps. This skill guides Claude, Codex, or Claude Code through a focused security review.

Firestore rules can accidentally allow data leaks, role escalation, or unsafe updates. This skill guides Claude, Codex, and Claude Code through a focused review of access control and validation gaps.

Mobile teams need safe ways to change app behavior without shipping a new build. This skill guides Remote Config setup, defaults, template edits, and reviewed CLI deployments.