Coding standards, frameworks, refactoring, and code review (18.5K skills)
Use this skill to keep code changes focused and consistent with the project’s existing patterns. It helps size a plan, compare smaller alternatives, reuse established mechanisms, and simplify a diff before handoff or while reviewing agent-written code.
Investigate suspected bugs in the Octane monorepo by reproducing and narrowing the failure, locating the owning code, and adding a regression test. It also helps determine whether behavior that differs from React is an intentional design choice.
Create a new Octane component in a `.tsrx` file, convert JSX into `.tsrx`, or add scoped styles and themes. The skill directs the agent to the authoring rules and a nearby example so new files follow the project’s conventions.
Triage an Octane GitHub issue from its number or link. The skill gathers issue context, identifies its type and affected project area, checks relevant rules, and develops a proposed fix and validation plan, with implementation as an optional next step.
Audit Octane bindings and assess maintenance needs such as lifecycle fixes, dependency updates, metadata drift, or redundant copied files. For selected findings, make a bounded change backed by current evidence about source ownership, compatibility, and the binding’s supported behavior.
Use TanStack DB from Octane components to display live query results, handle loading and errors, paginate data, and pace mutations such as frequent text edits. The skill also covers query dependencies and live effects.
Port a React library into Octane from npm or GitHub inputs. The workflow checks dependencies and license evidence, implements the full published surface, and verifies tests, types, provenance, and package readiness.
Review Octane changes for risks to frequently executed code, including unstable V8 object shapes, forced DOM layout, and extra scheduling steps. The skill scans changes, checks findings against runtime behavior, and records the evidence needed before a pull request is ready.
Use this skill to prepare an Octane change for a pull request. It guides branch and worktree setup, focused commits, validation, review gates, and a draft PR body that records agent provenance and relevant evidence.
Use Codex CLI to inspect code, refactor it, or make requested edits. The skill guides command setup, sandbox selection, and session resumption, then helps review Codex’s output critically so you can decide what to do next.
Analyze slow interactions, choppy scrolling, and unresponsive pages with browser measurements. The skill breaks down input delay, processing, and rendering time, finds long animation frames and blocking scripts, and connects interaction timing with layout shifts.
Use a collection of JavaScript snippets to measure and investigate web performance in a browser. Choose checks for loading speed, Core Web Vitals, interactions, images, video, and network conditions, then interpret structured results and provide practical recommendations.
Assess a page’s connection type, bandwidth, latency, and data-saving preference to guide loading choices for different network conditions. The skill provides connection-aware budgets and practical options for adjusting images, video, scripts, and other resources.
Measure page loading and trace delays across server response, rendering, scripts, fonts, and resource hints. The skill uses follow-up checks to identify blocking assets and loading patterns, helping prioritize improvements to the initial page experience.
Measure LCP, CLS, and INP, three indicators of loading speed, visual stability, and interaction responsiveness. The skill guides follow-up checks based on results to help identify likely causes of slow rendering, layout shifts, or delayed interactions.
Prepare and triage GitHub issues using repository evidence. The skill checks contribution rules, issue forms, labels, and existing reports, then helps shape a suitable issue or route a confirmed duplicate appropriately.
Prepare a pull request candidate by checking its approved issue link, branch naming, template sections, type label, and verification evidence. The skill helps contributors assemble an accurate review package while keeping remote actions separately authorized.
Audit existing AI agent skills for metadata, triggers, structure, length, actionable instructions, decision points, output requirements, and references. Get a severity-grouped report with proposed changes; apply safe edits only when explicitly requested.
Guide defect fixes from an approved issue and a reproducible problem on the current main branch. Map affected flows, tests, evidence, change size, and rollback boundaries, then report validation results while keeping review evidence separate from delivery authority.
Apply a structured harness for substantial coding work: clarify scope and acceptance criteria, track large tasks, use test-first development when meaningful tests apply, and delegate or investigate when useful. It also sets boundaries for review workload and risky changes.
Plan commits around complete, reviewable units of work. Keep related tests and documentation with each change, and use those units to organize larger implementations into chained pull requests when appropriate.
Plan a reviewable delivery for oversized changes by identifying cohesive work units and arranging them as chained pull requests or one PR. The skill tracks dependencies, review scope, line counts, and verification needs while following the chosen delivery strategy.
Create or update reusable AI agent skills with clear triggers, concise runtime instructions, and valid frontmatter. Get a structured SKILL.md plus optional assets or references, with guidance on avoiding duplicates and refreshing the skill registry.
Work with an agent to clarify requirements, explore distinct solution options, and compare how well each option meets the needs. The process can carry a selected, detailed approach into demoable implementation slices, with related planning documents kept in sync.