Guides for Claude Code, Codex and OpenCode

Updated 24 October 2026

Setup guides and fixes for the coding agents Agent Pets works with. Each one gives the exact settings, file paths and commands, with copy-paste snippets checked against the agent’s own documentation.

Claude Code

• Claude Code notifications: when done or waiting for input: Claude Code notifications when it’s done, waiting for input or needs permission: macOS alerts, Ghostty, iTerm2, VS Code, hooks, and answering from Chrome.
• Claude Code notifications not working? Fixes: Claude Code notification late, missing or hook not firing? The six-second rule, unsupported terminals, /hooks, macOS and Linux settings, tmux, and questions that never alert.
• Claude Code sound when done: play a sound when it finishes or needs you: Claude Code sound when done: a notification sound or beep when it finishes or needs input, with hooks for macOS, Linux, Windows and WSL, plus spoken alerts.
• Claude Code notifications on your phone (ntfy, Pushover, Telegram): Claude Code notifications on your phone with ntfy, Pushover or Telegram, named by project and only when you’re away, and how to approve from your phone.
• Claude Code notifications on Windows and WSL: Claude Code alerts on Windows: Windows Terminal bell and taskbar flash, PowerShell toast hooks, BurntToast, WSL via powershell.exe, VS Code, and fixes.
• Claude Code multiple sessions: run and track agents in parallel: Run Claude Code multiple sessions in parallel and see which needs you: agent view, worktrees, alerts that name the project, and an agent dashboard in Chrome.
• Run Claude Code agents in parallel with git worktrees: Run Claude Code agents in parallel without collisions: one worktree per session with claude --worktree or git worktree add, .env, ports, merging and cleanup.
• Using Claude Code in tmux: Claude Code in tmux: notifications with allow-passthrough, Shift+Enter, scrollback and mouse, truecolor, a window per project, and a hook that flags the window.
• Claude Code hooks: a practical guide: Claude Code hooks explained: every hook event, where hooks live, matchers, stdin JSON, exit codes, JSON decisions, debugging, and 8 copy-paste recipes.
• Claude Code permissions explained: How Claude Code permissions work: permission modes, Shift+Tab, allow/ask/deny rules and their syntax, settings precedence, extra directories and sandboxing.
• How to stop Claude Code asking for permission: allow all commands safely: Stop Claude Code asking “Do you want to proceed?”: auto-accept edits, an allow list, auto mode, how to allow all commands, and why it still asks.
• Claude Code YOLO mode (--dangerously-skip-permissions): what it does and safer options: Claude Code YOLO mode is --dangerously-skip-permissions: what it skips, the command, VS Code, switching mid session, the risks and safer options.
• Claude Code settings.json: file locations and reference: Where Claude Code’s settings.json lives on macOS, Linux and Windows, which file wins, the keys worth knowing, env vars, /config, and common mistakes.
• Claude Code statusline: setup and examples: Set up a Claude Code statusline: the /statusline command, the statusLine setting, the JSON it gets, example scripts, and what to do when it’s not showing.

Codex

• Codex notifications: get notified when Codex is done: Get a Codex notification when it’s done or needs approval: built-in alerts, a notification sound, a notify program, hooks, and fixes when nothing shows up.
• Codex approval modes: full auto, YOLO mode and approval_policy: Codex approval_policy and sandbox_mode explained: what replaced full auto, what YOLO mode does, never vs on-request, untrusted, /permissions and profiles.
• Codex config.toml: location, example and options: Where Codex config.toml lives (~/.codex, CODEX_HOME, Windows, project), a full example, the options worth setting, API keys, profiles, -c overrides and fixes.
• Codex hooks: hooks.json guide: Codex hooks.json explained: where it lives, every event, matchers, stdin JSON, exit codes, trusting hooks with /hooks, notify vs hooks, and copy-paste recipes.

OpenCode

• How to get OpenCode notifications: Get notified when OpenCode finishes or needs permission: OpenCode 2's built-in alerts, notification plugins for OpenCode 1 and 2, and approving OpenCode from Chrome.
• OpenCode permissions config: allow all, ask, deny and examples: OpenCode permissions config and examples for v1 and v2: how to allow all, ask and deny rules, wildcards, rule order, per-agent rules and Allow always.
• OpenCode plugins: how to write and install one: Install OpenCode plugins from npm or a local folder, and write your own: Plugin.define, tool hooks, events, examples for OpenCode 2 and 1, and debugging.

Every agent

• Notifications for Claude Code, Codex, OpenCode and omp compared: What Claude Code, Codex, OpenCode and omp alert you about out of the box: desktop alerts, bells, hooks and plugins, approval events, and one setup for all four.
• Coding agent dashboard: Claude Code, Codex and OpenCode in one view: Six ways to watch many AI coding agents at once: Claude Code’s agent view, tmux, Herdr, OC-Claw, Agent Pets and a DIY hook dashboard, compared.

Comparisons

• Agent Pets vs Claude Code agent view: Claude Code agent view vs Agent Pets: a free terminal dashboard for background sessions vs pets in Chrome you can approve and reply from.
• Agent Pets vs Herdr: Herdr vs Agent Pets: a free terminal multiplexer for coding agents vs pets in Chrome you can approve and reply from. Use them together.
• Agent Pets vs OC-Claw: Agent Pets vs OC-Claw: a free open-source notch pet for many agents vs pets in Chrome you can approve and reply from. Honest comparison.