搜索文档、Skill 和 MCP
「扩展智能体」分类共 662 个 Skill，按仓库 star 排序。分类自动生成，仅供参考。
ancoleman/ai-design-components525Evaluate LLM systems using automated metrics, LLM-as-judge, and benchmarks. Use when testing prompt quality, validating RAG pipelines, measuring safety (hallucinations, bias), or comparing models for production deployment.
JetBrains/thinkrail514Use when composing an ask_user_question round inside a workflow, or when a workflow skill names it at a question step. Shared norms for the tool — not a workflow, nothing to execute.
skydoves/compose-performance-skills514Use this skill when a Compose tree uses SubcomposeLayout, BoxWithConstraints, or Scaffold and the developer reports extra measure passes, slow first frame, or layout passes running content composition repeatedly. Covers why SubcomposeLayout composes its slots during the measure phase, why BoxWithConstraints forces a subcomposition for every new Constraints value, why nesting Scaffold or BoxWithConstraints multiplies the cost, when a custom Layout or Modifier.layout { } replaces SubcomposeLayout, and how to use SubcomposeLayoutState's slot reuse policy and precompose APIs when SubcomposeLayout is genuinely required. Use when the developer mentions BoxWithConstraints, SubcomposeLayout, Scaffold, "extra measure pass", "double measurement", "first frame slow", "subcompose", or notices that wrapping content in BoxWithConstraints regresses scroll perf inside a LazyColumn.
skydoves/compose-performance-skills514Use this skill to decide when Jetpack Compose derivedStateOf is the right tool and when it is pure overhead. Covers the "input frequency must exceed output frequency" rule, the mandatory remember { derivedStateOf { } } wrapper, the canonical pitfall of capturing non-state variables by initial value (and the remember(key) fix), and the snapshotFlow alternative for fire-and-forget side effects on derived values. Use when the developer mentions derivedStateOf, scroll-position-driven booleans, threshold checks, firstVisibleItemIndex, "show FAB on scroll", recomposition counts that don't drop after wrapping a value, or asks whether a computed string concatenation should use derivedStateOf.
skydoves/compose-performance-skills514Use this skill to drive Compose HotSwan from an AI agent (Claude Code, Cursor, any MCP client) so the agent can edit a Kotlin file, trigger a hot reload, capture a device screenshot, evaluate the result against a design intent, and iterate without a human in the loop. Covers the seven HotSwan MCP tools (hotswan_get_status, hotswan_reload, hotswan_take_screenshot, hotswan_start_snapshot, hotswan_stop_snapshot, hotswan_select_variant, hotswan_build_and_install), the canonical edit-reload-screenshot loop, snapshot-based rollback, and when to fall back to a full install for schema changes. Use when the developer says "get the AI to tune this screen until it matches a mock", asks "can the AI see what changed?" or "can the AI screenshot the device?", sets up a Claude Code or Cursor workflow that needs MCP tool access, or wants AI-driven UI iteration.
HRuiCcc/RuiC-card-skill514Generate interactive 3D holographic collectible-card websites from a user description or reference image, using layered artwork, Blender and Three.js. Includes project-local Blender installation, reusable parallax materials and browser verification. Host-agnostic — any harness that runs a multimodal model can use it, since the model draws the layer artwork and inspects the rendered frames. Use when the user asks for 全息闪卡, 镭射卡, 3D 卡牌网页, a holographic collectible card site, or an editable card.blend.
skydoves/compose-performance-skills514Use this skill to fix unstable Jetpack Compose types once a stability diagnosis has identified them. Covers the three-tier strategy — make the type truly stable with val plus immutable fields, mark with @Immutable or @Stable when the source is owned, and use stabilityConfigurationFiles for third-party or Java types. Explains the compiler-level difference between @Immutable and @Stable (static expression promotion), kotlinx.collections.immutable for List/Set/Map parameters, and the StableHolder wrapper escape hatch. Use when the developer asks how to stabilize a User class, a List parameter, java.time.LocalDateTime, a Flow parameter, or when the compiler report shows unstable params and the developer wants the fix. The diagnostic step lives in a sibling skill.
skydoves/compose-performance-skills514Use this skill to explain why the Compose compiler classified a class or composable parameter as stable, runtime, unknown, or unstable. Covers the 12-phase inference algorithm, the five compiler-level stability types (Certain / Runtime / Unknown / Parameter / Combined), the generic bitmask encoding (Pair=0b11, ImmutableList=0b1), the Known Stable Constructs registry, and the runtime `$stable: Int` field generated by `@StabilityInferred`. Use when the developer asks "why is X classified as Y?", when a stability report shows a surprising `runtime stable`, `unknown`, or `unstable` verdict, when generics, inheritance, cycles, interfaces, or cross-module classes are involved, or when the user mentions `$stable`, `@StabilityInferred`, separate compilation, or "the compiler thinks my class is unstable but it looks fine".
skydoves/compose-performance-skills514Use this skill to install and operate the `skydoves/compose-stability-analyzer` IntelliJ / Android Studio plugin so the developer sees Compose stability feedback live in the editor instead of waiting for a Gradle build. Covers installing the plugin from disk, configuring the `Settings → Tools → Compose Stability Analyzer` panel, reading the four gutter colors (green stable, red unstable, yellow runtime, gray no-params), the per-parameter hover documentation, the inline parameter hint badges, and the `UnstableComposable` weak-warning inspection with its `@Suppress("NonSkippableComposable")` and `@Suppress("ParamsComparedByRef")` quick fixes. Use when the user mentions gutter icons, inline hints, the stability inspection, the IDEA plugin, real-time stability feedback while editing, or asks why a composable is flagged as non-skippable in the editor.
coco-research/coco511Train and optimize AI agents using Microsoft's Agent Lightning framework with reinforcement learning. Use when setting up agent training, instrumenting agents with tracing, configuring LightningStore, implementing reward functions, or optimizing prompts with RL/APO algorithms.
coco-research/coco511Post-run self-evaluation system that scores agent output on correctness, clarity, actionability, and conciseness. Use after /team runs, skill executions, or when explicitly asked to evaluate output quality.
coco-research/coco511Wrap any command-line tool into a JSON-emitting agent skill. Use when you want to make a CLI reliably callable and parseable by an AI agent — introspect its --help, define a structured-output (--json) contract and a stable error contract, then generate a SKILL.md wrapper with verified examples. Triggers on "wrap this CLI", "make X agent-usable", "generate a skill for this command", "turn a tool into an agent skill".
coco-research/coco511Audit token consumption across agents, skills, MCP servers, and rules. Identifies bloat, redundant components, and produces prioritized token-savings recommendations. Use when the context window is filling up too fast or before adding new components.
coco-research/coco511Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.