Loading...
Loading...
Found 1,641 Skills
Behavioral guardrails for Cavekit agents. Four principles — think before coding, simplicity first, surgical changes, goal-driven execution — that prevent over-engineering, silent assumptions, scope creep, and unfocused work. Every task-builder, reviewer, planner, and inspector must internalize these before writing a single line. Trigger phrases: "guardrails", "karpathy", "scope creep", "over-engineering", "stop adding features", "surgical fix".
Lovrabet Runtime CLI — Manage application directories, dataset queries, data CRUD, SQL execution, and BFF invocations via the lovrabet command. Trigger words: Cloud Diagram, lovrabet, lovrabet-cli, app list, dataset, data filter, data getOne, create, update, delete, sql exec, bff exec, accessKey, compress, jq.
Build AI agent interfaces with Polpo UI — composable React chat components, CLI tools, and starter templates. Use when the user wants to create a chat app, add chat components, install @polpo-ai/chat, scaffold a Polpo project, configure theming/dark mode, use ChatInput, ChatMessage, ChatSessionList, or any Polpo UI component. Triggers on "polpo ui", "chat UI", "chat component", "@polpo-ai/chat", "@polpo-ai/ui", "create-polpo-app", "chat input", "session list", "agent selector", "chat interface", "polpo chat", "chat widget", "multi-agent".
Access and navigate GenomeArk AWS S3 bucket - VGP assemblies, QC data, and species directory structure
Interact with KWeaver Knowledge Network and Decision Agent — build knowledge networks, query Schema/instances, semantic search, execute Action, Agent CRUD and conversation, Trace data analysis. Interact with Dataflow document processes — list processes, trigger runs, query run history, view step logs. Interact with Skill management module — register Skill, search in market, progressive reading, download and installation. Interact with Toolbox / Tool — create toolbox, upload OpenAPI tools, publish, start and stop. Interact with Vega observability platform — query Catalog/resources/connector types, health inspection. This skill is automatically activated when users mention intents such as "knowledge network", "knowledge graph", "query object type", "execute Action", "what Agents are there", "create Agent", "converse with Agent", "list all Agent templates", "list Agents I created", "list Agents in private space", "dataflow", "data flow", "process orchestration", "process run records", "process logs", "trigger dataflow", "view dataflow run history", "Skill", "skill package", "register Skill", "install Skill", "read SKILL.md", "toolbox", "toolbox", "upload tool", "register tool", "OpenAPI tool", "enable tool", "publish toolbox", "data source", "data view", "atomic view", "Catalog", "Vega", "health check", "inspection", "trace", "evidence chain", "data flow tracking", "data source", "how data is obtained", etc.
Compact SGLang torch-profiler triage skill. Use when Codex should inspect an existing `trace.json(.gz)` or profile directory, trigger `sglang.profiler` against a live server, and return one compact report with kernel, overlap-opportunity, and fuse-pattern tables. Single-trace triage is enough for quick diagnosis; mapping+formal two-trace triage gives stronger overlap conclusions.
One-stop skill for the project architecture center — draft new architecture documents, refresh existing architecture documents, or conduct an architecture health check. Automatically determine the mode based on user input: `new` (draft)/ `update` (refresh to the latest code status)/ `check` (view only, generate issue list). The `check` mode has three sub-goals: consistency within a single feature design, alignment between design and code, and consistency among multiple documents under `easysdd/architecture/`. Single-target rule — only modify one document or check one target at a time. Trigger scenarios: User says "fill in an architecture doc", "draft an architecture document", "refresh the architecture directory", "write down the structure of this module", "conduct an architecture check", "is the design internally consistent?", "does the plan match the code?", "are there conflicts among several documents in the architecture folder?", or when it is found in the feature-design / feature-acceptance / implement phase that an architecture action needs to be performed first before proceeding.
Draft or update requirement documents under `easysdd/requirements/` for the project — describe a capability's "reason for existence, solution approach, and boundaries" using **user stories + plain language**, so non-technical readers can quickly grasp the key highlights of the system. Layered with architecture: requirement is the "problem space" (why this capability is needed), while architecture is the "solution space" (what structure is used to implement it). Two modes: new (draft a new requirement doc from scratch), update (refresh an existing doc based on new materials or implementation changes). Single-target rule — only modify one document at a time. Trigger scenarios: when the user says "fill in a requirement doc", "write down the requirements for this capability", "update the requirements directory", or when it is found during the feature-design phase that there is no corresponding requirement for the capability to be implemented this time.
Write or update external guide documents for the project —— dev-guide (for contributors/integrators/downstream developers) and user-guide (for end users). The output is stored in the project's docs/ directory, maintained alongside the code, and searchable by search tools. Difference from libdoc: guidedoc is task-oriented ("How to do Y with X"), while libdoc is reference-oriented ("What each part of X looks like"). Trigger scenarios: When the user says "write documentation", "developer guide", "user guide", or proactively push at the end of feature-acceptance.
One-stop skill for the project architecture center — draft new architecture documents, refresh existing ones, or conduct an architecture health check. Automatically determine the mode based on user input: `new` (draft)/ `update` (refresh to latest code status)/ `check` (review without modification, generate issue list). The `check` mode has three sub-objectives: consistency within a single feature design, alignment between design and code, and consistency among multiple documents under `codestable/architecture/`. Single-target rule — only modify one document or check one target at a time. Trigger scenarios: User says "fill in an architecture doc", "draft an architecture document", "refresh the architecture directory", "write down this module structure", "conduct an architecture check", "is the design internally consistent?", "does the plan match the code?", "are there conflicts among several documents in the architecture folder?", or when an architecture action is required before proceeding during the feature-design / feature-acceptance / implement phases.
Dodun 操作手册。Use when the user asks how Dodun works, how to install or use the CLI, how to join a space, send or receive with a token, manage skills, inspect inbox or attachments, manually route a task, or debug why routing / execution / receive-back failed. Prefer this over the old `dodun-agent-network` wording and only use legacy register/search/send commands when the user explicitly wants low-level directory debugging.
Analyze A-share stocks using fixed scripts, supporting the maintenance of local watchlists and position pools, fetching individual stock and concept sector data, querying risks such as financial report disclosures and share reduction announcements, and outputting structured recommendations including Buy, Watch, Hold, Reduce Position, and Sell. Suitable for scenarios where you need to update local stock pools, fetch real-time data, compare the top three strongest stocks in a concept sector, or generate transaction analysis with risk prompts.