All Skills

Query Polymarket: markets, prices, orderbooks, history.

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

Manim CE animations: 3Blue1Brown math/algo videos.

vLLM: high-throughput LLM serving, OpenAI API, quantization.

Search personal knowledge bases, notes, docs, and meeting transcripts locally using qmd — a hybrid retrieval engine with BM25, vector search, and LLM reranking. Supports CLI and MCP integration.

HuggingFace hf CLI: search/download/upload models, datasets.

Scan codebases for known vulnerabilities using OSV, CVE, and GHSA databases.

Delegate coding work to Codex, Claude Code, OpenCode, or Pi as background workers; not simple edits or read-only code lookup.

Design one-off HTML artifacts (landing, deck, prototype).

Local speech-to-text with the Whisper CLI (no API key).

阶段④ 执行。主 agent 跟踪进度 + 并行派生子 agent 实现模块，含代码门禁与调试档位。

阶段② 详细设计。根据 design.md 第一章补全第二章各模块设计 + 第四章 ADR。

Use when the team needs API routes, action contracts, request and response shapes, error handling, pagination, or webhook patterns for a backend or BFF. Trigger on requests to design APIs before implementation.

Use when debugging slow SQL queries, designing database indexes, or rewriting queries for better performance. For EXPLAIN analysis, index strategy, N+1 problem resolution, and query optimization patterns.

Use when the team needs to understand how data moves through frontend, backend, database, jobs, auth, and external integrations. Trigger on requests to map data flow, event flow, or state transitions across the system.

Use this when background workers or job processors need design or review. Trigger on requests about asynchronous processing, job payloads, retries, scheduling, worker ownership, or long-running tasks outside the request path.

Use this when the team needs to improve latency, throughput, render performance, query speed, worker throughput, or resource usage across frontend, backend, database, or background systems.

Use this when frontend or service code needs help integrating with an API, provider, or internal backend contract. Trigger on requests about request flow, response handling, auth propagation, retries, or client-side integration patterns.

Use when the team needs sign-in methods, profiles, sessions, invitations, memberships, or auth-related architecture for a Supabase app. Trigger on requests to design authentication and authorization using Supabase Auth.

Use this when the system needs clear analysis of trusted versus untrusted execution contexts, external inputs, client/server boundaries, provider callbacks, or admin-level code paths. Trigger on requests about trust boundaries, attack surfaces, or where verification must happen.

Use this when the project needs Python-specific backend architecture for APIs, services, workers, scripts, or internal tools. Trigger on requests about Python runtime structure, framework fit, module boundaries, or service organization.

Use when a Firebase app needs Firestore collection design, document shape, subcollection strategy, security rule implications, or query-aware data modeling. Trigger on requests to design Firestore data structures before implementation.

Use when the team needs to decide how frontend state should be organized across React or Next.js apps. Trigger on requests about local state, global state, server cache, forms, optimistic updates, or choosing between context, stores, and query-based patterns.

Use this when planning environment variables, config files, secrets, flags, and provider settings across local, CI, staging, production, serverless, or multi-runtime systems.

Use this when a project needs a better local development workflow for setup, running services, seeding data, testing integrations, or managing multi-service developer environments.

Use when the team needs a top-level system architecture spanning frontend, backend, auth, data stores, jobs, integrations, and deployment boundaries. Trigger on requests to design the overall system or validate a proposed architecture.

Use this when the project needs a clear strategy for local, test, staging, production, cloud, serverless, or hybrid environments. Trigger on requests about environment layout, runtime differences, environment-specific behavior, or deployment-context planning.

Use this when a frontend feature needs a clear plan for server data fetching, caching, invalidation, mutations, optimistic UI, or auth-aware data loading.

Use this when designing Python scripts for automation, batch processing, file operations, data transformation, or internal workflows that need deterministic structure and safe environment handling.

Design API contracts from product flows and backend responsibilities. Use when the team needs resource models, action endpoints, request and response shapes, error handling, pagination rules, and contract clarity before implementation begins.

Use when the team needs a clear product point of view, product principles, or a concise philosophy statement before planning UX, features, or architecture. Trigger on requests about product vision, differentiation, or what the product should optimize for.

You are a data triager that answers data questions by routing them across a network of skills and agents. These questions may involve pulling data from a Snowflake warehouse, computing metrics, locating relevant tables or fields, diagnosing a metric change, or performing exploratory analysis. Your job is to determine how to answer a question and which skills and subagents to invoke.

Use this skill when creating or updating PR descriptions. Provides detailed guidelines for writing clear, comprehensive pull request descriptions with proper Summary and Testing sections.

Centralized 'Truth Engine' for cross-jurisdictional legal context (US, EU, CA) and contract scaffolding.

Anti-over-engineering guardrail that activates when an AI coding agent expands scope, adds abstractions, or changes files the user did not request.

First-principles assumption auditor. Classifies each hidden assumption (fact / convention / belief / interest-driven), ranks by fragility × impact, and rebuilds conclusions from verified premises. Bilingual: auto-detects Chinese or English.