josephyaduvanshi

Use when delegating a coding, refactoring, review, or research task to an external AI coding-agent CLI (Codex, Gemini, Qwen, OpenClaude, OpenCode) or to a Claude subagent, or when running multiple agents in parallel. Routes to the right delegation mechanism and the correct per-agent driver skill.

Use when delegating a task to a separate Claude Code process via the `claude` CLI (print mode or interactive tmux) — e.g. headless runs in an isolated worktree, a different account/settings, or nested orchestration. For ordinary Claude subagents prefer the built-in Agent tool. Load delegate-to-agents first.

Use when delegating a coding or research task to the Google Gemini CLI (`gemini`) from Claude Code. Covers non-interactive `-p`, approval modes, output formats, sessions, worktrees, and tmux interactive driving. Load delegate-to-agents first for the shared mechanism.

Use when delegating a coding, refactoring, or code-review task to the OpenAI Codex CLI (`codex`) from Claude Code. Covers non-interactive `codex exec`, sandbox/approval flags, interactive tmux driving, resume, review, and parallel worktrees. Load delegate-to-agents first for the shared mechanism.

Use when delegating a coding task to the OpenClaude CLI (`openclaude`) from Claude Code. OpenClaude is a Claude-Code-compatible agent CLI whose standout feature is --provider (anthropic/openai/gemini/github/bedrock/vertex/ollama). Covers print mode, permission modes, providers, and tmux. Load delegate-to-agents first.

Use when delegating a coding task to the OpenCode CLI (`opencode`) from Claude Code. Covers one-shot `opencode run`, interactive TUI via tmux, sessions, PR review, and parallel work. NOTE opencode is not installed by default on this machine. Load delegate-to-agents first.

Use when delegating a coding or research task to the Qwen Code CLI (`qwen`) from Claude Code. Covers one-shot positional prompts, approval modes, sessions (and the --chat-recording gotcha), output formats, and tmux interactive driving. Load delegate-to-agents first for the shared mechanism.

Use when an external agent (Codex/Gemini/Qwen/OpenClaude/OpenCode) produces a patch you must review, test, and accept/reject before keeping — isolated git worktree, untrusted-diff reconciliation, independent test run, and accept/partial/reject handoff. Ported and generalized from the hermes-agent kanban-codex-lane skill.

Internal helper contract for calling the qwen-companion runtime from Claude Code

Internal guidance for composing Qwen Code / Qwen 3.x prompts for coding, review, diagnosis, and research tasks inside the Qwen Claude Code plugin

Internal guidance for presenting Qwen helper output back to the user

Internal helper contract for calling the gemini-companion runtime from Claude Code

Internal guidance for composing Gemini 2.5 Pro/Flash prompts for coding, review, diagnosis, and research tasks inside the Gemini Claude Code plugin

Internal guidance for presenting Gemini helper output back to the user

Expert Zustand (v5) state management for React — writing stores, selectors, middleware (persist/devtools/immer), TypeScript typing, Next.js/SSR architecture, testing, and eliminating unnecessary re-renders. Use this skill whenever the user creates or edits a Zustand store, mentions zustand, create()/createStore/useStore/useShallow/persist, asks about React global/client state management or state libraries, hits "Maximum update depth exceeded" / "getSnapshot should be cached" errors, complains about React re-renders tied to a store, migrates from Redux/Context to Zustand or from Zustand v4 to v5, or wires state into Next.js App Router — even if they don't say "zustand" explicitly but the code imports it. Also covers state management in Tauri desktop/mobile apps (Rust tauri::State vs webview stores, invoke/events, multi-window sync, tauri-plugin-store persistence) — use it for any Tauri + React state question, and for "where should this state live" decisions in Tauri apps generally.