System Prompts
Establishes an AI's identity as an insightful question interpreter, guiding it to deeply understand inputs and provide concise, bulleted answers.
Defines an AI's role to analyse content and apply relevant, predefined tags, outputting a structured JSON object for metadata management.
Deconstructs AI prompts, extracting identity, purpose, steps, and output instructions. Structures the analysis into a detailed Markdown format for clarity.
Extracts world model and task algorithm updates from content, identifying underlying beliefs and procedural knowledge for decision-making and task performance.
Analyses tool documentation to generate clear, structured instructions, including an overview, usage syntax, common use cases, and key features.
Evaluates content's 'wow-factor' by measuring surprise, novelty, insight, value, and wisdom per minute, providing a comprehensive quality score.
Defines Greybeard, a principal systems engineer, to generate secure, reliable, and auditable system prompts with robust injection resistance and privacy preservation.
Provides expert strategies and tactics for optimising Large Language Model prompts, enhancing clarity, relevance, and output quality.
Evaluates Honeycomb queries against specified language info and guidelines, providing a detailed critique and a pass/fail judgment for quality assurance.
Classifies and rates content based on predefined labels, thematic relevance to human meaning and AI impact, and idea density, providing a quality score.
Offers a structured template for crafting detailed AI prompts, outlining sections for defining identity, goals, steps, and desired output for optimal performance.
Assesses AI response quality against human expert performance, providing a letter grade, detailed reasons, and a 1-100 score for comprehensive evaluation.
Assesses the quality of AI/ML/LLM work results using a comprehensive, multi-dimensional rating system, providing a numerical score and corresponding human-level evaluation.
Establishes a universal AI identity focused on deeply understanding input and generating the best possible, most relevant output.
Defines an expert AI assistant for the Fabric framework, suggesting appropriate patterns and commands based on user requests to optimise AI integration.
Defines the methodology for an AI to concisely summarise other prompts. It outlines a structured approach for generating brief, active-voice summaries, focusing on purpose, nuance, and expected output.
Claude Code agent prompt intended for CLI-based software engineering tasks. Focuses on completing coding tasks efficiently with concise responses, excels in task management, git operations, and defensive security practices.
Claude Code agent prompt intended for CLI-based software engineering tasks. Focuses on completing coding tasks with concise responses under 4 lines, excels in task management with TodoWrite, code conventions following, and defensive security practices.
Claude web automation agent prompt intended for browser-based tasks with strong security controls. Focuses on safely completing web automation tasks while defending against prompt injection attacks, excels in strict verification requirements, privacy protection, and copyright compliance.
Claude Sonnet 4.5 general-purpose assistant prompt intended for conversational AI interactions. Focuses on answering questions, creative writing, and analysis with balanced conciseness, excels in web search integration, artifact creation, and thorough responses to complex topics.
Augment Code agent prompt intended for IDE-integrated coding assistance using Claude Sonnet 4. Focuses on understanding codebases through context engine retrieval and executing coding tasks, excels in task management, package manager usage, and safe file editing with str_replace operations.
GPT-5 agent prompt intended for an agentic coding assistant with read/write access to a developer codebase. Focuses on orchestrating safe code edits and targeted information gathering, excels in tasklist-guided planning, concise Markdown output and cautious tool use.
Cluely assistant prompt intended for general problem-solving and screen analysis. Focuses on providing specific, accurate, and actionable responses without meta-phrases, excels in technical problem-solving with fully commented code and LaTeX math rendering.
Cluely enterprise prompt intended for live-meeting co-pilot assistance during conversations. Focuses on answering questions directly, providing term definitions, and advancing conversations with follow-up questions, excels in structured bullet-point responses, objection handling, and concise format without headers or pronouns.
CodeBuddy chat assistant prompt intended for conversational interaction in chat mode. Focuses on answering questions, providing explanations, and engaging in natural conversation, excels in quick direct responses and asking clarifying questions without waiting for all information.
CodeBuddy craft mode prompt intended for software engineering tasks with full tool access. Focuses on accomplishing coding tasks through iterative tool use and step-by-step execution, excels in file editing, command execution, and maintaining concise responses while following strict workflow rules.
Comet Assistant prompt intended for web automation and information retrieval tasks. Focuses on finding information and answering user queries through browser control tools, excels in thorough task completion, strict safety rules, citation requirements, and protection against prompt injection attacks.
Cursor AI coding assistant prompt intended for interactive CLI-based software engineering tasks powered by GPT-5. Focuses on autonomously resolving coding tasks following user instructions, excels in parallel tool execution, code style compliance, and providing concise status updates with structured summaries.
Cursor AI coding assistant prompt intended for IDE-based pair programming tasks powered by GPT-4.1. Focuses on autonomously resolving coding tasks following user instructions, excels in semantic code search, comprehensive context understanding, task management with todo tracking, and precise code editing with citation support."
Cursor AI coding assistant prompt intended for IDE-based pair programming tasks powered by GPT-5. Focuses on autonomously resolving coding tasks with structured todo tracking and status updates, excels in parallel tool execution, semantic code search, and providing concise progress summaries with proper code citations.
Cursor AI coding assistant prompt intended for IDE-based pair programming tasks powered by Claude Sonnet 4. Focuses on following user instructions and resolving coding tasks autonomously, excels in parallel tool execution, semantic code search, and precise code editing with line-based citation support.
Cursor AI coding assistant prompt intended for IDE-based pair programming tasks powered by GPT-4.1. Focuses on autonomously resolving coding tasks following user instructions, excels in semantic code search, comprehensive context understanding, task management with todo tracking, and precise code editing with citation support.
Cursor AI coding assistant prompt intended for IDE-based conversational assistance powered by GPT-4o. Focuses on answering coding questions and providing explanations, excels in semantic code search, file reading, and concise responses with line-based code citations.
Devin AI assistant prompt intended for codebase question-answering tasks. Focuses on finding relevant code and answering queries with precise citations, excels in mermaid diagram creation, matching user language, and providing structured responses with file path references.
Devin AI software engineer prompt intended for autonomous coding tasks with full tool access. Focuses on understanding codebases, implementing features, and fixing bugs, excels in shell commands, code editing, semantic search, browser automation, and deployment operations.
Dia AI assistant prompt intended for web browser-based conversational assistance inside the Dia browser. Focuses on answering questions with structured responses, excels in media integration including images and videos, Ask Dia hyperlinks for follow-up questions, and rich formatting with Simple Answers, tables, and LaTeX support.
Emergent E1 agent prompt intended for full-stack application development in React/FastAPI/MongoDB Docker environments. Focuses on building launchable MVPs efficiently through frontend-first development with mock data, excels in shadcn UI components, backend integration, automated testing workflows, and integration with third-party services.
Google Antigravity fast prompt intended for rapid coding assistance in the Google Gemini IDE on Windows. Focuses on quick code generation and debugging with concise responses, excels in pair programming workflows, web application development with HTML/CSS/JavaScript/Next.js/Vite, and implementing rich aesthetics with modern design principles.
<identity> You are Antigravity, a powerful agentic AI coding assistant designed by the Google Deepmind team working on Advanced Agentic Coding. You are pair programming with a USER to solve their coding task. The task may require creating a new codebase, modifying or debugging an existing codebase,…
SPECIAL INSTRUCTION: think silently if needed Act as a world-class senior frontend React engineer with deep expertise in Gemini API and UI/UX design. Using the user's request, your primary goal is to generate complete and functional React web application code using Tailwind for excellent visual aest…
Junie AI assistant prompt intended for read-only code exploration and information retrieval tasks. Focuses on quickly investigating project structures and retrieving relevant code snippets, excels in fuzzy search, file structure analysis, and special commands like search_project, get_file_structure, and answer for comprehensive responses.
You are an intent classifier for a language model. Your job is to classify the user's intent based on their conversation history into one of two main categories:
System Prompt Identity You are Kiro, an AI assistant and IDE built to assist developers. When users ask about Kiro, respond with information about yourself in first person.
Identity You are Kiro, an AI assistant and IDE built to assist developers. When users ask about Kiro, respond with information about yourself in first person.
You are Leap, an expert AI assistant and exceptional senior software developer with vast knowledge of REST API backend development, TypeScript and Encore.ts.
Lovable AI assistant prompt intended for real-time web application development through a chat interface. Focuses on creating and modifying React/Vite/Tailwind/TypeScript applications with beautiful responsive designs, excels in Supabase integration, image generation, SEO best practices, and efficient code editing workflows.
Manus AI agent prompt intended for general-purpose task automation in a Linux sandbox environment. Focuses on information gathering, data analysis, research writing, and web development through iterative tool use, excels in browser automation, file operations, shell commands, and deploying websites with public access."
Manus AI agent prompt with detailed modules for task automation in Ubuntu sandbox environment. Focuses on structured task planning through planner, knowledge, and datasource modules with event stream processing, excels in todo tracking, data API integration, and detailed research writing in prose format.
Manus AI assistant prompt intended for general-purpose task automation with detailed capability documentation. Focuses on information processing, content creation, problem-solving, and web development tasks, excels in browser automation, file operations, shell commands, and deploying websites with public access.
Notion AI assistant prompt intended for workspace management and content creation inside the Notion productivity app. Focuses on answering questions, searching workspace content, and creating or editing pages and databases, excels in Notion-flavored markdown formatting, semantic search across connected sources, and SQLite queries for structured data analysis.
Bolt AI coding assistant prompt intended for in-browser web application development using WebContainer Node.js runtime. Focuses on creating full-stack applications with React, Vite, and Supabase integration, excels in artifact-based code generation, database migrations with RLS policies, and starting development servers without verbose explanations.
Cline AI coding assistant prompt intended for autonomous software engineering tasks in IDE environments. Focuses on completing coding tasks step-by-step with file editing and command execution, excels in search-based code analysis, code definition exploration, MCP server integration, and browser automation for web testing.
OpenAI Codex CLI prompt intended for terminal-based coding assistance with a concise, friendly personality. Focuses on completing coding tasks through terminal commands and patch-based file editing with update_plan tracking, excels in working with sandbox configurations, approval workflows, and providing structured final answers with minimal verbosity.
OpenAI Codex CLI prompt intended for terminal-based agentic coding assistance with git-backed workspace and rollback support. Focuses on completing coding tasks by applying patches and running shell commands, excels in sandboxed execution, pre-commit checks, and providing friendly teammate-like responses for codebase questions.
Google Gemini CLI prompt intended for interactive terminal-based software engineering assistance. Focuses on understanding codebases, planning implementations, and executing coding tasks with self-verification loops, excels in following project conventions, running build/lint checks, and delivering concise responses under 3 lines with extreme conciseness.
Proton Lumo AI assistant prompt intended for conversational AI interactions across multiple platforms including mobile and web. Focuses on providing thoughtful, balanced analysis with multiple perspectives while maintaining intellectual honesty, excels in handling sensitive topics educationally, file analysis for CSV/PDF/code, and recommending Proton ecosystem products.
RooCode AI coding assistant prompt intended for IDE-based software engineering tasks in VSCode. Focuses on completing coding tasks with minimal code changes and maintainability focus, excels in multiple modes including Code, Architect, Ask, Debug, and Boomerang for workflow orchestration.
Knowledge cutoff: 2024-06 <rules> - Identify if the user request is about cloning a website based on the conditions provided in the cloning_instructions. - If the user request is not a cloning request, invoke `generate_design_system` if you find the user request relevant. If the query is too vague o…
You are a powerful agentic AI coding assistant called Orchids working with a Next.js 15 + Shadcn/UI TypeScript project. Your job is to follow the user's instructions denoted by the <user_query> tag.
Perplexity AI search assistant prompt intended for answering user queries using web search results. Focuses on writing accurate, detailed, and comprehensive answers with proper citations, excels in well-formatted markdown output, handling multiple query types including academic research, news, coding, and creative writing.
This advanced execution agent operates within a multi-agent system, completing tasks for Poke while maintaining strict separation from user interaction. It excels at parallel task orchestration, precise context handling, and trigger management, ensuring reliable results for delegated goals. Designed for expert system integration and robust workflow automation.
Poke AI assistant prompt intended for conversational interaction via text messages on iMessage, WhatsApp, and SMS. Focuses on interpreting user messages and delegating tasks to execution agents, excels in witty and warm personality, adaptive texting style matching, and managing automations without revealing agent architecture to users.
Poke AI assistant supplementary prompt intended for handling WhatsApp messaging platform limitations. Focuses on working within 24-hour message windows and template restrictions, excels in emoji reaction interpretation for yes/no confirmations and gracefully managing platform constraints.
Poke AI assistant supplementary prompt intended for handling error scenarios, user frustration, and agent communication workflows. Focuses on maintaining the unified entity illusion while handling mistakes, excels in email and calendar draft confirmation workflows, custom MCP integration handling, and proper agent message routing.
Poke AI assistant supplementary prompt intended for handling integrations, user onboarding context, and membership policies. Focuses on routing queries to appropriate data sources like Notion, Linear, and email, excels in providing information about enterprise security, privacy features, and membership pricing while maintaining a chill, human-like personality.
Poke AI assistant supplementary prompt intended for handling email link formatting, notification protocols, and memory systems. Focuses on proper markdown link formatting with emoji shortlinks, concise email notification summaries with actionable links, and context preservation without explicitly referencing memory construction.
Poke AI assistant supplementary prompt intended for handling long conversation context and user memory persistence. Focuses on maintaining conversation continuity through automatic summarisation, excels in remembering user preferences and communication patterns without explicitly mentioning memory access to users.
Qoder AI coding assistant prompt intended for IDE-based pair programming tasks. Focuses on following user instructions to solve coding tasks including debugging, code creation, and question answering, excels in task planning with detailed steps, mandatory code validation after changes, parallel tool execution for read operations, and memory management for project-specific knowledge.
Qoder AI background agent prompt intended for autonomous task execution based on design documents. Focuses on creating actionable implementation plans from design specifications without user interaction, excels in breaking down tasks into verifiable steps, code validation, and providing brief completion summaries.
Qoder AI design specialist prompt intended for creating technical design documentation from user ideas. Focuses on guiding users through design creation with repository analysis and iterative feedback, excels in generating structured design documents with UML diagrams, Mermaid charts, and industry-standard templates for various project types.
Replit AI coding assistant prompt intended for the Replit online IDE environment. Focuses on assisting users with coding tasks including file modifications, shell commands, and natural language questions, excels in proposing file changes, package installations, workflow configurations, deployment setup, and database management within the Replit Linux and Nix environment."
Same.dev AI coding assistant prompt intended for the Same cloud-based IDE at same.new. Focuses on pair programming with users to develop web applications with autonomous task completion, excels in parallel tool execution, shadcn/ui customization, website cloning, automatic deployment to Netlify, and version management within a Docker Ubuntu environment."
Trae AI builder prompt intended for IDE-based pair programming tasks in the Trae AI IDE. Focuses on following user instructions to create, modify, or debug codebases, excels in task management with todo tracking, code citation with special XML formats, web search citations, and step-by-step code change implementation with minimal tool calls.
Trae AI chat prompt intended for IDE-based conversational interactions in the Trae AI IDE using the AI Flow paradigm. Focuses on analyzing user tasks and determining when tool usage is required, excels in code editing with precise placeholder notation, web citation formatting, code reference tracking, and structured todo list management.
Traycer AI phase mode prompt intended for IDE-based task decomposition in the Traycer AI IDE. Focuses on breaking down user queries into high-level executable phases that maintain compilation and test integrity, excels in codebase exploration via search tools, clarification gathering, and creating phased implementation plans with referred file references.
v0 prompt intended for web-based React and Next.js code generation on the v0.app platform. Focuses on creating full-stack applications using CodeProject blocks with best practices, excels in Tailwind CSS design implementation, shadcn/ui component integration, structured task management, and seamless integrations with Supabase, Neon, Stripe, and other services.
VSCode Agent chat title prompt intended for generating brief conversation titles within VSCode. Focuses on creating pithy titles of 8 words or fewer that capture the main topic, excels in following Microsoft content policies, avoiding copyrighted content, and producing short impersonal responses.
VSCode Agent Claude Sonnet 4 prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on following user requirements carefully and completing tasks through tool usage, excels in expert-level knowledge across programming languages, creative problem solving, workspace exploration, and producing complete runnable solutions with proper testing.
VSCode Agent Gemini 2.5 Pro prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on following user requirements carefully and completing tasks through tool usage, excels in expert-level knowledge across programming languages, creative problem solving, workspace exploration, and producing complete runnable solutions with beautiful modern UI for webapps.
VSCode Agent GPT-4.1 prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on completing tasks until fully resolved and taking proactive action without unnecessary questions, excels in expert-level knowledge across programming languages, creative problem solving, apply_patch tool for efficient file edits, and producing complete runnable solutions with beautiful modern UI for webapps.
VSCode Agent GPT-4o prompt intended for IDE-based coding assistance in VS Code. Focuses on following user requirements carefully and completing tasks through tool usage, excels in expert-level knowledge across programming languages, creative problem solving, workspace exploration, replace_string_in_file tool for file edits, and producing complete runnable solutions with beautiful modern UI for webapps.
VSCode Agent GPT-5 Mini prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on completing tasks end-to-end with structured todo list tracking and quality gates, excels in expert-level programming knowledge, friendly conversational tone, requirements coverage tracking, apply_patch tool for efficient edits, and proactive extras with validation before completion.
VSCode Agent GPT-5 prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on completing tasks end-to-end with structured todo list tracking and quality gates, excels in expert-level programming knowledge, friendly conversational tone, requirements coverage tracking, apply_patch tool for efficient edits, and proactive extras with validation before completion.
VSCode Agent NES tab completion prompt intended for real-time inline code completion in VS Code. Focuses on predicting and completing code changes based on recently viewed snippets, edit history, and cursor position, excels in maintaining code consistency, understanding developer patterns, and providing relevant suggestions within special edit tags.
VSCode Agent general prompt intended for IDE-based coding assistance in VS Code on Windows. Focuses on answering user requests using relevant tools and following requirements carefully, excels in expert-level knowledge across programming languages, semantic code search, file editing with insert_edit_into_file tool, terminal command execution, error validation, and project creation including VS Code extensions and Jupyter notebooks.
Warp.dev Agent Mode prompt intended for terminal-based software development assistance in the Warp AI terminal. Focuses on helping with development questions and tasks through terminal commands, excels in command execution, file editing with edit_files tool, grep search for code exploration, version control integration, and handling secrets securely with environment variables.
Windsurf Cascade prompt intended for IDE-based pair programming in the Windsurf IDE on Windows. Focuses on solving coding tasks independently and collaboratively using the AI Flow paradigm, excels in persistent memory system for context retention, proactive codebase research, structured planning with update_plan tool, and safe terminal command execution with browser preview capabilities.
Windsurf Cascade tools prompt intended for providing tool definitions and capabilities to the Cascade AI assistant in the Windsurf IDE. Focuses on defining available actions for browser preview, code editing, file operations, terminal execution, and memory management, excels in structured tool schema definitions with clear descriptions and parameter specifications.
Xcode DocumentAction prompt intended for generating documentation for Swift code within the Xcode IDE. Focuses on understanding the selected code and file context to produce accurate documentation, excels in Swift-specific documentation generation with access to full file contents and selected code snippets.
Xcode ExplainAction prompt intended for explaining selected Swift code within the Xcode IDE. Focuses on providing detailed explanations of code functionality and behaviour, excels in breaking down complex Swift code with references to the full file context.
Xcode MessageAction prompt intended for generating commit messages or code-related messages within the Xcode IDE. Focuses on analysing selected code changes to produce appropriate messaging, excels in generating context-aware messages based on the file and code context.
Xcode PlaygroundAction prompt intended for generating Swift Playground code within the Xcode IDE. Focuses on creating interactive Swift code examples and experiments, excels in generating runnable playground code with access to full file and selected code context.
Xcode system prompt intended for an editor or Xcode session analysing Apple platform Swift projects. Focuses on identifying missing types, guiding safe code changes and giving concise, actionable advice, excels in favouring Swift/Apple APIs, cautious verification and tasklist-guided searches.
Z.ai Code agent prompt intended for interactive CLI or IDE sessions developing Next.js 15 TypeScript projects. Focuses on building production-ready, frontend-first applications while enforcing the prescribed stack and dev rules, excels in z-ai-web-dev-sdk integration, Prisma usage and project-aware tooling.
Acts as a dedicated prompt engineer in a conversational refinement loop: each turn produces a revised prompt alongside targeted clarifying questions, continuing until the user confirms the result meets their needs.