Where to start
- Setting up a hook, a skill or a plugin? Each one has its own guide, with the steps and the file it needs.
- A setting you cannot find? The settings.json guide lists all four files and says which one wins.
- Want Claude Code to follow your rules? Memory, hooks and skills are the three roads, and each has a guide here.
- Not sure why a feature is not showing up? Several of these guides start from that exact problem.
Every guide in this topic
- AGENTS.md in Claude Code: What Claude Actually Reads
Claude Code has read AGENTS.md on its own since v2.1.277, and skips it the moment a CLAUDE.md sits at or above your working directory.
- How to Add an MCP Server to Claude Code
The exact claude mcp add commands for HTTP, SSE and local servers, what each scope stores and where, and how to check, list and remove one.
- Claude Code Bedrock and Vertex setup: env vars and gateway
Set up Claude Code on Amazon Bedrock or Google Vertex AI: the environment variables, model pins, credentials, a corporate gateway, and what breaks.
- Claude Code channels: push events into a session
Channels push chat and CI events into an open Claude Code session. How to turn one on, who can use them, and the reasons it will not register.
- Claude Code rewind: checkpoints, undo and what it misses
Undo Claude Code work with /rewind, /checkpoint or /undo: restore code, conversation or both from a checkpoint, and see what a rewind cannot undo.
- CLAUDE.md and Claude Code memory: what loads, and when
Where Claude Code looks for CLAUDE.md, what belongs in it, and the four reasons it seems to ignore the file you wrote.
- Claude Code code review: which command, and what it finds
Which Claude Code review command to run, what each one costs, what a review finds and misses, and how to review every pull request.
- Claude Code commands cheat sheet: every slash command
Every built-in Claude Code slash command in one grouped list, the CLI flags worth typing, and the keys that save the most time.
- Claude Code diff panel: what /diff shows and needs
Claude Code's /diff panel shows Claude's edits beside the conversation. What it needs, the keys it answers to, and why it can show nothing.
- Claude Code effort levels: auto, max and ultracode
What each Claude Code effort level changes, how auto resolves, where /effort and CLAUDE_CODE_EFFORT_LEVEL fit, and what ultracode adds.
- Claude Code environment variables: what each one changes
Claude Code environment variables that change how it runs, grouped by job, where each one belongs, and why a value set in the wrong place looks ignored.
- Claude Code headless mode: claude -p in scripts and CI
Run Claude Code without a terminal: the print-mode flags, output formats, permission settings for scripts and CI, and why a run hangs or prints nothing.
- Claude Code Hooks: What They Are and How to Set One Up
Every Claude Code hook event, where hooks are configured, three copy-paste examples, the exit-code and JSON rules, and what the docs warn about.
- Claude Code keyboard shortcuts and keybindings.json
Every Claude Code keyboard shortcut by context, plus the keybindings.json file: the actions you can bind, null to unbind, and why a rebind does nothing.
- Claude Code with local models and LLM gateways
Run Claude Code on Ollama, OpenRouter or LiteLLM: the variables a gateway needs, what stops working without an error, and the limit on your subscription.
- Claude Code Mods: What They Are and How to Build One
A Claude Code mod is a plugin whose hooks run inside the app, and some of them draw. What a mod is, and how it differs from hooks and plugins.
- Claude Code notifications: bell, desktop alerts and hooks
How Claude Code tells you a task finished or needs you: the bell and desktop alert setting, the Notification and Stop hooks, and why a hook stays silent.
- Claude Code OpenTelemetry: metrics, events and traces
Turn on Claude Code's OpenTelemetry export: the variables, where they must live, what leaves the process, how to redact it, and how to prove it works.
- Claude Code Permission Modes, Explained
What each Claude Code permission mode allows without asking, what no mode auto-approves, and how Omniscio's Agent Permission Level sits beside them.
- Claude Code plan mode: how it works and when to use it
Plan mode is Claude Code's read-only planning phase. How to enter and leave it, which model plans, and what to do when it edits files anyway.
- Claude Code plugins: what they are, and how to install one
Install a Claude Code plugin from a marketplace, choose the right scope, and read the install summary so you know what reached your session.
- Claude Code Prompt Suggestions: Turn Off or Get Them Back
The gray suggested prompt, the settings that turn it off, which switch actually wins, and why it stops appearing when you want it back again.
- Claude Code sandbox: what it blocks and how to turn it on
Turn on Claude Code's sandbox with /sandbox or a settings key, see what it confines, and fix the git, gh and network failures it causes.
- Where Is Claude Code settings.json? All 4 Paths, Which Wins
~/.claude/settings.json, the project .claude/settings.json, settings.local.json and managed-settings.json: where each one lives and which one wins.
- Claude Code Skills: What They Are and How to Add One
A Claude Code skill is a folder with a SKILL.md that loads only when relevant. How to write one, and how it relates to slash commands and AGENTS.md.
- Claude Code voice dictation: /voice, hold mode and tap mode
Voice dictation in Claude Code: turn on /voice, choose hold or tap mode, rebind the key, set the language, and fix the messages that stop it.
Set it up once, then run it in every session
Omniscio is the desktop app we make for people who run several Claude Code sessions at once. Any model, any harness — Claude Code, Codex, OpenCode, Devin, Pi, and Hermes all supported.