docs(command-development): add SlashCommand tool coverage

[sjnims]

Summary

The official Claude Code docs have a detailed section on the SlashCommand tool which can programmatically invoke custom commands. The command-development skill doesn't cover this feature.

What's Missing

1. SlashCommand Tool Overview

The tool allows Claude to execute custom slash commands programmatically during conversations. Users should understand:

• What it does
• How commands become available to it
• How to control its behavior

2. disable-model-invocation Context

The skill documents this frontmatter field but doesn't explain its relationship to the SlashCommand tool:

disable-model-invocation: true  # Prevents SlashCommand tool from calling this command

3. Permission Rules

Official docs show permission patterns:

• Exact match: SlashCommand:/commit
• Prefix match: SlashCommand:/review-pr:*
• Deny all: Add SlashCommand to deny rules

4. Character Budget

The SlashCommand tool has a character budget (default 15,000) limiting command descriptions shown to Claude. This affects:

• How many commands Claude "sees"
• Why keeping descriptions concise matters
• The SLASH_COMMAND_TOOL_CHAR_BUDGET env var

5. Requirements for SlashCommand Visibility

Commands must have description frontmatter field to be available to SlashCommand tool.

Suggested Changes

Add a new section to SKILL.md or create references/slashcommand-tool.md covering:

• What the SlashCommand tool is
• How to make commands available/unavailable to it
• Permission configuration
• Character budget considerations
• Best practices for programmatic invocation

References

• Official docs: https://code.claude.com/docs/en/slash-commands.md#slashcommand-tool

[sjnims]

PR submitted: #56