Skip to main content

​
Overview

Maestro’s slash command system lets you create reusable prompts that are sent to AI agents with a simple /command syntax. Commands support template variables that are dynamically replaced at runtime, making them adaptable to different contexts.

​
Command Types

​
Built-in Maestro Commands

Handled internally by Maestro (not sent to the agent):
CommandDescription
/historyGenerate synopsis and add to History panel
/wizardStart Auto Run planning wizard
/skillsList Claude Code skills (Claude Code only)

​
Custom AI Commands

User-defined commands that send prompts to the agent with variable substitution.

​
Agent Native Commands

Provider-specific commands discovered automatically (e.g., Claude Code’s /compact, /cost). See Slash Commands for complete documentation.

​
Creating a Custom Command

1

Open command editor

  1. Press Cmd+, (Mac) or Ctrl+, (Windows/Linux) to open Settings
  2. Navigate to AI Commands tab
  3. Click Add Command
2

Configure the command

Fill in the command details:
  • Trigger: The slash command (e.g., /deploy)
  • Description: What the command does
  • Prompt: The text sent to the agent (supports template variables)
3

Save and test

  1. Click Save
  2. Type the trigger in any AI tab
  3. Verify autocomplete shows your command
  4. Press Tab or Enter to insert and execute

​
Template Variables

Variables are replaced with actual values when the command is executed:

​
Agent & Session Variables

# Deploy Command Example

Deploy the application at {{AGENT_PATH}} to production.
Agent: {{AGENT_NAME}}
Branch: {{GIT_BRANCH}}
Session: {{AGENT_SESSION_ID}}
Available variables: 
{{AGENT_NAME}}       - Agent display name
{{AGENT_PATH}}       - Full path to agent's working directory
{{AGENT_GROUP}}      - Group name (if grouped)
{{AGENT_SESSION_ID}} - Provider session ID
{{TAB_NAME}}         - Custom tab name
{{TOOL_TYPE}}        - Agent type (claude-code, codex, opencode, factory-droid)
See the templateVariables.ts implementation for the complete list.

​
Conductor Profile Variable

{{CONDUCTOR_PROFILE}} inserts your “About Me” profile from Settings → General: 
Remember my preferences: {{CONDUCTOR_PROFILE}}

Now help me with this task: ...
This is useful for mid-conversation reminders about your communication style or technical context.

​
Command Argument Patterns

​
Explicit Placement with $ARGUMENTS

Control exactly where user input appears: 
# /plan command
Create a detailed implementation plan for: $ARGUMENTS

Include:
- Technical approach
- Dependencies
- Testing strategy
Usage: /plan user authentication with OAuth2 Result: 
Create a detailed implementation plan for: user authentication with OAuth2

Include:
- Technical approach
- Dependencies
- Testing strategy

​
Automatic Appending

If $ARGUMENTS is omitted, user input is appended: 
# /commit command
Please commit all changes with a descriptive message.
Usage: /commit fix login bug Result: 
Please commit all changes with a descriptive message.

fix login bug
Use $ARGUMENTS for precise control over prompt structure. Omit it for simple commands where appending is sufficient.

​
Real-World Command Examples

​
Daily Standup

Trigger: /standup
Description: Generate daily standup summary

Prompt:
It's {{WEEKDAY}}, {{DATE}}. I'm working on branch {{GIT_BRANCH}} at {{AGENT_PATH}}.

My profile: {{CONDUCTOR_PROFILE}}

Summarize:
1. What I worked on yesterday (check git log)
2. What I'm working on today
3. Any blockers or concerns

​
Code Review Request

Trigger: /review
Description: Request code review for staged changes

Prompt:
Review the staged changes in this repository.

Context:
- Branch: {{GIT_BRANCH}}
- Project: {{AGENT_PATH}}
- Session: {{AGENT_SESSION_ID}}

Provide:
- Security concerns
- Performance implications
- Code quality issues
- Testing recommendations

$ARGUMENTS
Usage: /review Focus on the authentication module

​
Documentation Generator

Trigger: /docs
Description: Generate documentation for code

Prompt:
Generate comprehensive documentation for: $ARGUMENTS

Format: Markdown compatible with our docs site
Include: Examples, parameters, return values, and edge cases
Style: {{CONDUCTOR_PROFILE}}

​
Deployment Checklist

Trigger: /predeploy
Description: Pre-deployment validation checklist

Prompt:
Running pre-deployment checks for {{AGENT_PATH}} on branch {{GIT_BRANCH}}.
Date: {{DATETIME}}

Validate:
- [ ] All tests pass
- [ ] No console errors or warnings
- [ ] Dependencies are up to date
- [ ] Environment variables are set
- [ ] Database migrations are applied
- [ ] Documentation is updated

Report any issues found.

​
Context-Aware Commit

Trigger: /commitcontext
Description: Commit with full context awareness

Prompt:
Create a commit for the staged changes.

Context:
- Project: {{AGENT_NAME}}
- Path: {{AGENT_PATH}}
- Branch: {{GIT_BRANCH}}
- Date: {{DATE}}
- Context usage: {{CONTEXT_USAGE}}%

Commit message should:
- Follow conventional commits format
- Reference any related issues
- Explain the "why" not just the "what"

Additional notes: $ARGUMENTS

​
Auto Run Integration

Commands can reference Auto Run context: 
Trigger: /loopreport
Description: Generate loop iteration report

Prompt:
Auto Run Loop Report

Document: {{DOCUMENT_NAME}}
Path: {{DOCUMENT_PATH}}
Iteration: {{LOOP_NUMBER}}
Timestamp: {{DATETIME}}

Summarize:
- Tasks completed this iteration
- Any errors encountered
- Metrics and performance
- Recommendations for next iteration
This command is useful when running in loop mode to track progress across iterations.

​
Spec-Kit and OpenSpec Commands

Maestro bundles structured workflow commands:

​
Spec-Kit Commands

From GitHub’s spec-kit: 
/speckit.help          - Learn spec-kit workflow
/speckit.constitution  - Define project principles
/speckit.specify       - Create feature specification
/speckit.clarify       - Identify gaps and ask questions
/speckit.plan          - Generate implementation plan
/speckit.tasks         - Create actionable tasks
/speckit.implement     - Convert to Auto Run documents
Prompts are stored in src/prompts/speckit/ and are editable.

​
OpenSpec Commands

From Fission’s OpenSpec: 
/openspec.help         - Learn OpenSpec workflow
/openspec.proposal     - Create change proposal
/openspec.apply        - Implement approved proposal
/openspec.archive      - Archive completed change
/openspec.implement    - Convert to Auto Run documents
Prompts are stored in src/prompts/openspec/.
Spec-Kit and OpenSpec commands are system commands and cannot be edited or deleted through the UI. Modify the source .md files to customize them.

​
Agent Native Command Discovery

Claude Code native commands are discovered automatically:
  1. When a Claude Code agent starts, Maestro spawns it with /help
  2. The system/init message contains all available slash commands
  3. Commands appear in autocomplete with “Claude Code command” label
Supported native commands in batch mode: 
/compact         - Compact conversation history
/cost            - Show token usage and cost
/init            - Initialize CLAUDE.md file
/pr-comments     - Address PR review comments
/release-notes   - Generate release notes
/review          - Request code review
/security-review - Perform security review
See Slash Commands for limitations.

​
Command Management

​
Editing Commands

  1. Open Settings → AI Commands
  2. Click the command to edit
  3. Modify trigger, description, or prompt
  4. Click Save

​
Deleting Commands

  1. Open Settings → AI Commands
  2. Click the command
  3. Click Delete
  4. Confirm deletion

​
Exporting/Importing Commands

Commands are stored in Maestro settings (settings.json). To share commands:
  1. Export settings via Settings → General → Storage Location
  2. Share the settings file with team members
  3. Or manually copy the customCommands array from settings JSON

​
Best Practices

/deploy, /review, /test are better than /d, /r, /t. Prioritize clarity over brevity.
Always provide context like project path, branch, date. The more context, the better the AI’s response.
Use {{GIT_BRANCH}}, {{DATE}}, {{AGENT_PATH}} to make commands reusable across projects.
If using $ARGUMENTS, test both with and without user input to ensure sensible defaults.
Add detailed descriptions so team members understand what each command does.

​
Troubleshooting

Command doesn’t appear in autocomplete Ensure:
  • Trigger starts with /
  • No duplicate triggers exist
  • Settings were saved
  • Agent is active (autocomplete only works in active tabs)
Variables not replaced Check:
  • Variable names match exactly (case-sensitive)
  • Double curly braces are used: {{VARIABLE}} not {VARIABLE}
  • Variable is supported (see template variables list)
Arguments not inserted Verify:
  • $ARGUMENTS is uppercase (not $arguments)
  • If omitted, arguments are automatically appended

Build docs developers (and LLMs) love

Get started for freeTalk to us