发现 Skills

Hard-enforcement protocol for building, updating, debugging Windmill Cloud scripts and flows for Build n Bloom. Replaces n8n-build for all NEW automation work. Use when creating, updating, debugging, or scheduling any Windmill script. Workspace bnb. User Bnb. All variables at u/YourUsername/* via wmill.getVariable() — never Deno.env.get for user secrets.

Structured 4-phase design workflow for new Liquers features. Use when designing substantial features requiring careful architecture (new value types, command libraries, storage backends, UI components, API endpoints). Triggers on "design a new...", "plan implementation of...", "architect the...", or explicit phase requests ("start Phase 1 design", "review Phase 2"). NOT for trivial additions (single command, bug fix, doc update).

Analyze and optimize system prompts using a structured prompting guidelines framework — AI-powered analysis and rewriting. Use when a prompt needs improvement, experiment results show quality gaps, or you want a structured review of an existing system prompt. Do NOT use when production traces show failures (use analyze-trace-failures first to identify patterns). Do NOT use to build evaluators (use build-evaluator).

Connect to GCP. No fumbling, no guessing, no 10 attempts. Just follow these exact steps in order.

Autonomous research and experimentation loop inspired by Karpathy's autoresearch. Continuously optimizes any measurable metric through systematic experimentation: modify → measure → keep/discard → repeat. Works with web performance, ads, code quality, ML, or any custom domain. Deploys a 4-agent team (lead, researcher, executor, analyst) via Agent Teams.

Scaffold the project structure, install dependencies, and configure the development environment

Playwright E2E testing patterns para Drupal. Trigger: Cuando se escriben tests E2E - Page Objects, selectores, workflow MCP.

Use when building MCP (Model Context Protocol) servers to enable LLMs to interact with external services through well-designed tools. Triggers include requests to create MCP servers, integrate external APIs or services, or build MCP tools in Python (FastMCP) or Node/TypeScript (MCP SDK).

E2E testing for Clerk apps. Use with Playwright or Cypress for auth flow tests.

Complete browser automation with Playwright. Auto-detects dev servers, writes clean test scripts to /tmp. Test pages, fill forms, take screenshots, check responsive design, validate UX, test login flows, check links, automate any browser task. Use when user wants to test websites, automate browser interactions, validate web functionality, or perform any browser-based testing.

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

Configure django-allauth with headless API, MFA, social authentication, and CORS for React frontends. This skill should be used when setting up authentication for a new Django project or adding django-allauth to an existing project that needs a React frontend integration. (project)

Execute Claude Code autonomously in the background for well-defined tasks, enabling parallel execution of up to 5 concurrent sessions while the user continues other work.

使用 git worktree + 背景代理会话管理大型任务（默认启动 Claude Code；可用 --codex 快速切到 Codex CLI，或通过 --agent-cmd 指定任意命令）。适合用户希望在不阻塞当前会话的情况下执行大型或多步骤任务。

ODIN's compact-form conversation skill — a formal-logic English register with predicate claims, Hoare-triple framing, and ASCII shortened-English keywords. Trigger when the user requests "axiom", "axiom-mode", "axiom-compact", or "compact form".

Mechanically tighten existing prose — restructure sections by dependency order, split or merge paragraphs, remove redundancy. Use to compress verbose plan files, READMEs, ADRs, and design docs. Does NOT change voice, register, tone, or any ODIN-mandated phrasing.

AG-UI protocol implementation guidance for event ordering (`RUN_STARTED`, `TOOL_CALL_*`, `STATE_SNAPSHOT`/`STATE_DELTA`), streaming semantics, and middleware patterns. Use when integrating agent backends with AG-UI clients.

Scan agent's session-history transcripts for save-worthy signals (corrections, preferences, decisions, references), propose and write auto-memory files with valid frontmatter and MEMORY.md entry. Use when the user says "save this to memory", "remember that", or "scan this session for memories".

Resolve code review comments by verifying their validity and proposing multiple solutions for confirmed issues. Use when addressing review feedback, analyzing whether review comments are valid, and generating architectural solutions (not naive fixes) for confirmed issues.

Install git pre-commit hooks via the project's hook tool of choice — Husky+lint-staged (npm/yarn/pnpm/bun), pre-commit framework (Python), lefthook (Go), cargo-husky (Rust), pre-commit (OCaml). Use when user wants commit-time formatting, linting, type-checking, or test gates. Detects ecosystem first; recommends the right tool.

Test-Driven Development (TDD) across any supported language. Use when implementing features or fixes with TDD methodology, writing tests before code, or following XP-style development.

Return the robot's own identity: name, owner, current preferences, learned facts about itself. Use whenever the agent needs to ground itself ('what's my name?', 'what do I prefer?') or to remind itself of its own evolving self-knowledge before responding.

Use when working with MySQL schemas before writing SQL or data-access code, mapping tables, columns, keys, indexes, entity relationships, inferring field meaning from metadata, or deciding whether sample data access needs user approval

Configure Fluentd and Fluent Bit for centralized log aggregation, routing, filtering, and enrichment across distributed infrastructure