LUDUS TECHNOLOGY
← All posts

AI slash commands and shorthands: the master blueprint

A comprehensive developer and creator guide to hidden shortcuts, semantic prompt hacks, and native integrations across ChatGPT, Claude, Gemini, Perplexity, and Grok.

How to read this guide. Most entries are phrases you type into the prompt and the model interprets, so the result depends on the model and its version. A few are real product features, which the blueprint labels as official: the Claude Code commands, the Gemini Workspace integrations and the Perplexity focus modes. I have not re-tested each entry against the vendors' current releases, so try one on a small task before you build a routine around it.
01 · OpenAI

ChatGPT

26 entries in 4 groups, listed as the blueprint gives them.

Visual & Structural Shorthands (Semantic Prompts)

  • /infographic

    Consolidates complex topics into structured, data-dense layout concepts combining text and visual metrics.

  • /cycle

    Maps out recurring workflows, iterative feedback loops, or circular systems step-by-step.

  • /diagram

    Draws high-level system architectures or relational concepts using text-based shapes or layout schemas.

  • /flowchart

    Breaks a multi-stage process down into sequential, conditional logic branches.

  • /mindmap

    Generates a conceptual tree branching outward from a central core theme for brainstorming.

  • /blueprint

    Generates highly detailed, technical, schematic-style layouts resembling architectural plans.

  • /xray

    Evaluates the hidden inner mechanics or deep conceptual layer behind a product or topic.

  • /explodedview

    Visually or structurally breaks an item apart to show how separate parts fit together.

  • /cutaway

    Renders a cross-section comparison showing the internal vs. external structural layers.

  • /stickynotes

    Brainstorms thoughts by placing exactly one concise point per simulated virtual colored note.

  • /handwritten

    Changes the tone and visual structure to resemble organic notebook sketches and handwritten prose.

Strategy & Execution Codes

  • /roadmap

    Formats a linear path of sequential execution milestones to reach an end goal.

  • /timeline

    Sequences historical data or chronological events in strict order of time.

  • /thenvsnow/beforeafter

    Renders a clean split-screen comparison of a core transition state.

  • /ecosystem

    Maps out every macro player, competitor, asset, and moving part within a specific industry.

  • /dashboard

    Formats summaries to resemble a software interface tracking high-level metrics and KPIs.

  • /swot

    Builds a classic four-quadrant analysis detailing Strengths, Weaknesses, Opportunities, and Threats.

Comprehension & Efficiency

  • /ELI5

    "Explain Like I'm 5." Translates complex corporate or scientific jargon into everyday analogies.

  • /TLDR

    "Too Long; Didn't Read." Ruthlessly strips massive documents down into 2-3 hyper-focused takeaways.

  • /BRIEFLY

    Applies an absolute token constraint, forcing the model to reply in 1-2 impactful sentences.

  • /EXEC SUMMARY

    Drafts a tight strategic briefing tailored for fast executive-level decision making.

  • /CHECKLIST

    Converts long instructional text into action-oriented, checkbox-style to-do tracking lists.

Native Interface & App Triggers

  • /canvas

    Opens a dedicated side-by-side editing canvas built for long-form copywriting or coding.

  • /search

    Forces an immediate deep web lookup before synthesizing an answer.

  • /deep_research

    Runs a multi-pass autonomous agent sequence for exhaustive topic synthesis.

  • /model

    Opens the modal selector to swap between active foundational engines mid-chat.

02 · Anthropic

Claude

8 entries in 2 groups, listed as the blueprint gives them.

Strategic Behavioral Shorthands

  • /godmode

    Instructs the model to drop passive corporate conversational fluff and prioritize intense, dense logical reasoning.

  • /ghost

    Activates highly organic, fluid human-sounding text optimization, bypassing typical robotic AI markers.

  • /10x

    Breaks an introductory concept or script code block out into a massive, highly scalable production framework.

  • /critique

    Overrides agreeableness, forcing Claude to aggressively tear down text and expose hidden logical fallacies.

  • /fable-prompter

    Acts as a meta-compiler, rebuilding messy user inputs into beautifully structured prompt engines.

Claude Code (Official Terminal Commands)

  • /plan

    Locks the environment into an analytical, read-only tracking state to map complex file changes before editing code.

  • /compact

    Condenses massive long-form chat logs to save context window space and reduce token overhead.

  • /doctor

    Automatically runs diagnostics across local system states, dependencies, and environment pipelines.

03 · Google

Gemini

8 entries in 2 groups, listed as the blueprint gives them.

Official Google Workspace `@` Integrations

  • @Gmail

    Hooks directly into your inbox to pull real-time dates, flight statuses, or message content.

  • @Google Docs

    Scans, references, or edits specific text documents saved inside your digital workspace.

  • @Google Drive

    Grants the model instant reading access across whole cloud folders and massive uploaded data PDFs.

  • @YouTube

    Forces the model to parse and summarize real video transcripts rather than searching web blogs.

  • @Google Maps

    Pulls location geometry, live routing, traffic density, and business metrics straight to your chat.

Gemini Custom Skills & Mental Shorthands

  • Truthmode

    Forces Gemini to evaluate your premises critically, highlighting flaws, biases, or logical errors.

  • 80/20

    Implements the Pareto Principle, ignoring tertiary advice to focus entirely on the top 20% high-impact actions.

  • Deepsearch

    Activates heavy multi-pass grounding loops across Google Search parameters before answering.

04 · Search Engine AI

Perplexity

6 entries in 2 groups, listed as the blueprint gives them.

Custom Macro Commands (User-Configured `/` Shortcuts)

  • /fact-check

    Restricts grounding algorithms entirely to academic repositories, clinical index sites, or verified data boards.

  • /list/table

    Reconfigures the system to bypass long introductory paragraphs, outputting straight tables or metrics.

  • /update

    Shunts standard indexing aside to scour the immediate last 24 hours of live global news wires.

Native Focus Modes (Hardwired Constraints)

  • Academic

    Locks search patterns exclusively onto peer-reviewed research papers and scientific databases.

  • Writing

    Cuts internet access entirely, instantly shifting the engine into a localized text generator that won't hallucinate external links.

  • Reddit / Social

    Forces search spiders to look exclusively inside community channels to track raw consumer sentiment.

05 · xAI

Grok

3 entries in 1 groups, listed as the blueprint gives them.

Real-Time Social & Analytical Modes

  • Fun Mode / Roast

    Grok's primary viral mode, tuning the system to reply with sharp sarcasm, humor, and direct roasts.

  • Regular Mode

    Balances output to deliver direct, neutral, corporate-appropriate informational summaries.

  • X-Grounding

    Instructs Grok to prioritize breaking news sentiment, raw video uploads, and trending discussions directly from the X platform ecosystem.

Start a conversation

Want a prompt playbook written around your own team?

Tell me which tasks eat your week and which tools your people already use. I will show you which of these shorthands are worth adopting and how to teach them.

Get in touch

Made with support of an LLM — by Ludus Technology Morocco.