839 skills · sorted by relevance
Audit website images and guide conversion to WebP and AVIF, formats that reduce image transfer size. Identify affected files, add fallback images, and review dimensions and loading behavior alongside the impact on initially visible content.
Check and fix hreflang tags, which identify language and regional versions of web pages for search engines. Review rendered HTML and HTTP responses to find missing reciprocal links, incorrect language codes, and fallback settings.
Review how a website behaves when images fail to load and add fallback images or meaningful placeholder content. Check loading behavior, consider loading placeholders, and use error logging and browser developer tools to investigate and verify fixes.
Inspect web font loading to identify invisible text, rendering delays, and layout shifts. Use measured evidence to guide fallback text display, preloading of important fonts, and the use of compressed formats such as WOFF2.
Review how an application displays prices, numbers, and dates, then replace manual formatting with Intl APIs that follow language and regional conventions. The skill also checks locale-aware sorting and fallback choices when a requested locale is unavailable.
Find broken or fragile image references in markup, styles, and components. Correct paths, restore missing assets, and add fallback images, then verify how the page behaves when image requests fail.
Build asset loading for Three.js scenes, including 3D models, textures, and HDR images for backgrounds and reflections. Coordinate multiple files, track loading progress, reuse cached assets, and handle failures with retries or fallback files.
Build and debug interactive web 3D scenes with Three.js, React Three Fiber, or Spline. Plan product configurators and portfolios, prepare models, connect scroll interactions, and review loading feedback, fallback content, accessibility, and performance on target devices.
Style materials with Anthropic's brand colors and typography. Apply distinct heading and body fonts, choose text colors for the background, and use coordinated accent colors for shapes, with fallback fonts when Poppins or Lora are unavailable.
Design and audit an agent's system prompt, the instructions that guide its work. Define a concrete outcome, starting input, available tools, completion checks, fallback behavior, and final report so the prompt describes a complete run.
Improve MNN inference on OpenCL by measuring bottlenecks and refining GPU kernels, the code that performs calculations. The skill guides baseline measurements, correctness checks, Android device benchmarks, and integration of new OpenCL features with compatibility fallbacks.
Review a data processing agreement against internal review standards, using the appropriate criteria for your company's role. Receive a clause-by-clause memo with risks, proposed edits, fallback positions, and inconsistencies with privacy policy commitments.
Run TypeScript checks that detect type errors without generating files. When shell execution fails because of environment problems or timeouts, switch to shell_agent to locate tools, address setup issues, and report the check results.
Build dashboard panels in plain TypeScript with data fetching and loading states. The pattern uses cached data, fallback values, and a circuit breaker, which temporarily pauses requests after repeated failures, to handle API problems.
Write large text files when write_file or shell_agent fails because of content size limits. The fallback passes multiline content through run_shell to Python, with guidance on escaping characters and checking the saved file.
Run TypeScript checks without generating output files, with a fallback when the usual commands fail. The workflow records existing errors, filters diagnostics to changed files, and repeats checks after fixes to assess the modified code.
Create typed data services for panels using direct or server-proxied API calls. Add caching and a circuit breaker, a mechanism that pauses requests after repeated failures, with cached or default data as fallback results.
Recover Python tasks when execute_code_sandbox fails with errors or timeouts. Install required packages, run the code through run_shell, and inspect messages and generated files to check the result.
Keep report writing focused on the original request when information retrieval fails. This skill guides agents to use existing knowledge, disclose missing information, create a structured document with write_file, and verify that the file exists.
Continue Python tasks when the sandbox cannot start by switching to run_shell. The workflow covers short scripts, saved script files, library checks, longer timeouts, and output verification for tasks such as PDF generation and data processing.
Run Python through run_shell when code execution tools repeatedly fail. This workflow embeds a multiline script directly in a shell command, supporting tasks such as CSV conversion and PDF generation while checking execution output.
Continue research when web-reading tools fail together by trying alternative URLs and using established knowledge where appropriate. Record inaccessible sources, mark references that need verification, and explain the limits of the resulting document.
Move Python execution to saved script files when execute_code_sandbox repeatedly fails or becomes unstable. The skill covers writing single or multiple files, running them through run_shell, inspecting outputs and errors, and removing temporary scripts afterward.
Run Python code directly on the host when repeated e2b sandbox failures block execution. The skill explains saving scripts, launching them through the shell, and checking outputs when an isolated environment is unnecessary.