Commands & Workflows

Commands (also called slash commands or workflows) are executable prompt files that AI IDEs can run on demand. They're different from rules/configurations — commands are action-based prompts you invoke with /command-name.

What are Commands?

While blueprints like AGENTS.md or .cursorrules define persistent rules and context for your AI assistant, commands are on-demand prompts that perform specific tasks when invoked.

Rules & Configs

Always active, define behavior and context. Examples: AGENTS.md,.cursorrules, CLAUDE.md

Commands

On-demand actions triggered by /command. Examples: security audit, code review, refactoring tasks

Supported Platforms

LynxPrompt supports commands across multiple AI IDEs. Each platform uses a different directory structure. Gemini CLI commands are TOML files; LynxPrompt converts the Markdown prompt for you. Devin Desktop still reads .windsurf/workflows/, and lynxp push picks those up too:

Platform	Directory	Invocation
Cursor	`.cursor/commands/*.md`	/command-name
Claude Code	`.claude/commands/*.md`	/command-name
Devin Desktop (formerly Windsurf)	`.devin/workflows/*.md`	/workflow-name
GitHub Copilot	`.github/prompts/*.prompt.md`	/prompt-name
OpenCode	`.opencode/commands/*.md`	/command-name
Gemini CLI	`.gemini/commands/*.toml`	/command-name
Kiro	`.kiro/prompts/*.md`	/prompt-name
Junie	`.junie/commands/*.md`	/command-name
Augment Code	`.augment/commands/*.md`	/command-name
Trae	`.trae/commands/*.md`	/command-name
Kilo Code	`.kilo/commands/*.md`	/command-name
Cline	`.clinerules/workflows/*.md`	/workflow-name.md
Roo Code / Zoo Code	`.roo/commands/*.md`	/command-name
Crush	`.crush/commands/*.md`	/command-name
Firebender	`.firebender/commands/*.mdc`	/command-name
Factory Droid	`.factory/commands/*.md`	/command-name
Qwen Code	`.qwen/commands/*.md`	/command-name
Pi	`.pi/prompts/*.md`	/prompt-name

Managing Commands with CLI

The LynxPrompt CLI automatically detects command files and lets you push, pull, and convert them between platforms.

Push a command to LynxPrompt

lynxp push .cursor/commands/security-audit.md

The CLI auto-detects this is a Cursor command and sets the type accordingly.

Convert between platforms

lynxp convert .cursor/commands/audit.md --target claude

Converts to .claude/commands/audit.md

Detect all commands in a project

lynxp detect

Scans your project for all configuration files and commands.

Creating Commands

Commands are Markdown files with instructions for the AI. They can include variables using [[VARIABLE]] syntax.

Example: Security Audit Command

# Security Audit

Perform a comprehensive security audit of the codebase.

## Tasks

1. Check for hardcoded secrets
2. Review authentication flows
3. Identify SQL injection risks
4. Check for XSS vulnerabilities
5. Review dependency versions

## Output

Provide a markdown report with:
- Severity levels (Critical, High, Medium, Low)
- File locations
- Recommended fixes

Finding Commands on Marketplace

Commands are available in the Blueprints marketplace. Use the type filter to show only commands:

All TypesAI Configs⚡ Commands

Command blueprints are marked with a ⚡ badge showing the target platform (e.g., "Cursor Command", "Claude Command").

Exporting Commands

When downloading a command blueprint, you can choose which platform format to export to. The download modal shows the target filename and directory.

Export to any supported platform

Download a Cursor command as a Claude command, a Devin Desktop workflow, a Gemini CLI TOML command, or any other format.

Ready to explore commands?

Browse community commands or upload your own.

Browse Commands