Custom Commands — Nyzhi Docs

Custom commands let you define reusable prompt templates as slash commands. They standardize common workflows like code review, test generation, or documentation. Share them with your team by committing .nyzhi/commands/ to version control.

Defining commands

Method 1: Markdown files

Create .md files in .nyzhi/commands/ in your project root:

<!-- .nyzhi/commands/review.md -->
# Review code for bugs and security issues
Review $ARGUMENTS for bugs, security vulnerabilities, and potential improvements.
Focus on edge cases and error handling.

The first line (# ...) becomes the command description
The rest is the prompt template
$ARGUMENTS is replaced with everything after the command name

Usage:

/review src/auth.rs

Expands to: “Review src/auth.rs for bugs, security vulnerabilities, and potential improvements. Focus on edge cases and error handling.”

Method 2: Config

Define commands inline in config.toml (see configuration):

[[agent.commands]]
name = "test"
prompt = "Write comprehensive tests for $ARGUMENTS. Cover edge cases, error paths, and happy paths."
description = "Generate tests for a module"

[[agent.commands]]
name = "explain"
prompt = "Explain how $ARGUMENTS works in detail. Include data flow, key types, and error handling."
description = "Explain a code path"

[[agent.commands]]
name = "doc"
prompt = "Write documentation for $ARGUMENTS. Include usage examples and parameter descriptions."
description = "Generate documentation"

Precedence

When the same command name exists in multiple sources:

Config commands override file-based commands with the same name
.nyzhi/commands/ takes precedence over .claude/commands/ (for compatibility)

$ARGUMENTS expansion

The $ARGUMENTS placeholder is replaced with everything the user types after the command name:

Input	Expansion
`/review src/main.rs`	`$ARGUMENTS` → `src/main.rs`
`/test the auth module`	`$ARGUMENTS` → `the auth module`
`/explain`	`$ARGUMENTS` → “ (empty string)

Listing commands

/commands

Shows all available custom commands with their descriptions.

Compatibility

Nyzhi also scans .claude/commands/ for command files, providing compatibility with Claude Code projects. If both directories contain a command with the same name, the .nyzhi/commands/ version wins.

Examples

Code review

<!-- .nyzhi/commands/review.md -->
# Thorough code review
Review $ARGUMENTS with focus on:
1. Correctness and edge cases
2. Security vulnerabilities
3. Performance issues
4. Error handling gaps
5. Test coverage

Present findings ordered by severity.

Refactor

<!-- .nyzhi/commands/refactor.md -->
# Safe refactoring
Refactor $ARGUMENTS to improve readability and maintainability.
- Keep changes minimal and reversible
- Maintain all existing behavior
- Update tests to match
- Run verification after changes

Documentation

<!-- .nyzhi/commands/doc.md -->
# Generate documentation
Write clear, concise documentation for $ARGUMENTS.
Include:
- Purpose and overview
- Public API with parameter descriptions
- Usage examples
- Error cases

Combine with skills and memory to teach the agent your project conventions.