Zest

CLAUDE CODE · COMMANDS

Claude Code commands
and slash commands, for teams.

Claude Code commands start with a slash: built-in ones like /init, /model and /permissions, and custom ones you write as SKILL.md files in .claude/skills/. Commit that folder and everyone on the repo gets the same commands (Anthropic's docs, October 2026).

Start free
npx @winding-labs/zest

Free for 3 seats · No credit card · One command signs in every agent on your machine

Sample data

Stack in the Zest app: how well your setup is wired, and what raises the score

USED BY ENGINEERING TEAMS AT

Flyps Charm Wildfires Geoforge MisaLabs

Claude Code slash commands that matter to a team

Type / to list the commands available to you, and keep typing to filter. A command runs only at the start of a message. Anthropic's reference lists more than a hundred. Some are bundled skills, such as /code-review and /loop: prompts Claude follows rather than fixed code. Not every command appears on every plan or platform. The ones a team leans on:

Built-in commands in Anthropic's commands reference, read October 5, 2026.
CommandWhat it does
/initWrites a starting CLAUDE.md for the project
/memoryEdits CLAUDE.md files and auto memory
/model, /effortSwitch the model (saved as your default) and the effort level
/permissionsManages the allow, ask and deny rules for tools
/usage (/cost, /stats)Session cost, plan usage limits and activity
/contextDraws context use as a grid, with what is filling it
/compact, /clearSummarize the conversation to free context, or start over empty
/planEnters plan mode
/resume, /rewindReopen an earlier session, or roll code and conversation back
/code-review (/review)Checks the diff, a branch or a pull request for bugs; --fix applies the findings
/security-reviewLooks for vulnerabilities in the branch's changes
/skills, /hooks, /mcp, /pluginShow what is loaded, and manage MCP servers and plugins
/doctorA setup checkup that finds configuration problems and offers fixes

Custom slash commands: write one as a skill

Custom commands and skills are one feature now. A file at .claude/commands/<name>.md still works, and a subfolder adds a prefix (frontend/component.md runs as /frontend:component). For new commands Anthropic suggests .claude/skills/<name>/SKILL.md: it can hold scripts and reference files, and it wins when both use the same name. A skill named like a built-in replaces it in the terminal, though the built-in's aliases still run the original.

This command fixes an issue by number. disable-model-invocation: true means it runs only when someone types it, and allowed-tools lets it use those gh and git commands without asking, for that turn:

.claude/skills/fix-issue/SKILL.md
---
description: Fix a GitHub issue by number, with a test
argument-hint: [issue-number]
disable-model-invocation: true
allowed-tools: Bash(gh issue view *) Bash(git add *) Bash(git commit *)
---

Fix GitHub issue #$0 following our coding standards:

1. Run gh issue view $0 and read the issue.
2. Write a failing test, then fix the code until it passes.
3. Commit with a message that references #$0.
Placeholders a command's body can use, from Anthropic's skills docs.
PlaceholderExpands to
$ARGUMENTSEverything typed after the command
$0, $1 (or $ARGUMENTS[0])One argument, counting from zero; quote a multi-word argument
$nameA named argument listed under arguments: in the frontmatter
${CLAUDE_SKILL_DIR}The skill's folder, for a bundled script
${CLAUDE_SESSION_ID}The current session's ID

How do I allow all commands in Claude Code?

Allow the commands you trust instead. A permission rule is a tool name with an optional pattern, and * matches any text. Claude Code checks deny rules first, then ask, then allow, and the first match wins however specific the others are. A bare Bash rule, or Bash(*), allows every shell command.

Rules from every settings file add up. /permissions edits them, --allowedTools sets them for one run, and answering a prompt with "Yes, and don't ask again" saves one to .claude/settings.local.json. A Bash rule matches the command's text, so git -C . push slips past Bash(git push *); use a hook when a rule must hold.

This rule set, from Anthropic's docs, runs npm scripts and commits without asking and blocks pushes and reads of .env. Permission modes, in the table below, go further; Shift+Tab cycles them in a session:

.claude/settings.json
{
  "permissions": {
    "allow": ["Bash(npm run *)", "Bash(git commit *)"],
    "deny": ["Bash(git push *)", "Read(./.env)"]
  }
}
Permission modes in Anthropic's docs. --permission-mode or defaultMode in settings picks the mode a session starts in.
ModeWhat it does
default (Manual)Asks the first time each tool is used
acceptEditsAlso edits files and runs common filesystem commands without asking
planReads and plans; doesn't edit source files
autoRuns without routine prompts while a classifier checks each action. The built-in starting mode in the terminal and VS Code; Anthropic says it reduces prompts but doesn't guarantee safety
dontAskDenies whatever would prompt: for CI and locked-down scripts
bypassPermissionsSkips prompts; --dangerously-skip-permissions turns it on

Is --dangerously-skip-permissions safe?

Only in a sandbox. Anthropic's docs say to use it only in isolated environments, such as containers, VMs or dev containers without internet access, where Claude Code can't damage your host, and note that it offers no protection against prompt injection. On Linux and macOS it refuses to run as root or under sudo.

A team can switch it off. Set permissions.disableBypassPermissionsMode to "disable" in managed settings to block bypass mode, and disableAutoMode does the same for auto mode. A repository's .claude/settings.json can't make either one the starting mode: Claude Code ignores those two values there.

Sharing commands across a team

Commit .claude/skills/ and any .claude/commands/ so everyone who clones the repository gets the same commands; the committed .claude/settings.json carries the team's permission rules and hooks with them. Claude Code reads skills in every parent folder up to the repository root, so a monorepo can keep shared commands at the top.

For more than one repository, ship the commands in a plugin, where they run as /plugin-name:command. Organization-wide skills can go in the managed settings directory. MCP servers add commands too: each MCP prompt shows up as /server:prompt.

Which commands your team actually runs

A command in the repository isn't one anybody uses. Zest's Claude Code plugin records every custom command and skill a session runs, by name, whether someone typed /name or Claude picked it. Built-in commands such as /clear and /compact, and Zest's own /zest: commands, are left out.

The AI Stack view lists each engineer's commands and skills next to their models and MCP tools, so a lead sees which ones spread across the team and which stay with their author. What else the plugin records is on Claude Code usage.

TRY IT ON YOUR TEAM

See what your team ships with AI.

Starter is free (3 seats, no credit card). Pro is $49 a month with 5 seats and Max is $149 a month with 10 seats — priced per workspace, not per engineer, with extra seats at $19 a month. See pricing

Start free
npx @winding-labs/zest

A FEW THINGS TO KNOW

Questions

Can a custom command run tools without asking?+

Yes. List them in the allowed-tools field of its frontmatter, such as Bash(git add *): Claude can use those tools without asking during the turn that runs the command. Your deny rules still apply.

Why does Claude Code say Unknown command?+

The command isn't available on your plan or platform, or was mistyped. For your own command, check that it is a folder with a SKILL.md inside and shows in /skills; a skill in a subfolder's .claude/skills appears only once Claude works on a file there.

Do Claude Code custom commands work in Codex?+

A skill does: Codex reads the same SKILL.md format from .agents/skills, and you run it as $skill-name. Files in .claude/commands are Claude Code's own format.