system-prompts-mcp-server

# Prompt Summaries ## Amp ### `claude-4-sonnet.yaml` - Positions the agent as Amp, a Sourcegraph-built coding assistant focused on end-to-end delivery of software tasks with balanced initiative. - Stresses heavy tool use (search, todo management, oracle) and mandatory diagnostics runs after code changes. - Enforces concise communication with markdown links to files and discourages unsolicited explanations. ### `gpt-5.yaml` - Shares Amp’s mission but tuned for GPT-5 with more aggressive expectations on parallel tool use and minimization of visible reasoning. - Adds strict formatting requirements (2–10 line final status with file:// links) and prioritizes automated planning via todo tracking and oracle/task tools. ## Anthropic ### `Claude Code 2.0.txt` - Configures Claude Code CLI with defensive-security-only scope, ultra-concise chat replies (≤4 lines), and mandatory TodoWrite planning for multi-step work. - Prioritizes tool-mediated context gathering, explicit code path references (`file:line`), and directs model to reference Claude Code docs via WebFetch when asked about platform features. ### `Sonnet 4.5 Prompt.txt` - Extends general Anthropic assistant behavior: conversational balance, intellectual curiosity, respect for cutoff, and empathy guidelines. - Details broad policy for math reasoning, citation practices, artifact usage, web search, and safety constraints for modern Claude deployments. ## Augment Code ### `claude-4-sonnet-agent-prompts.txt` - Defines Augment Agent persona with mandatory pre-edit retrieval, structured task management tooling, and XML snippet requirements for code excerpts. - Emphasizes conservative editing, package manager usage, and recovery strategies when stuck. ### `claude-4-sonnet-tools.json` - Enumerates Augment toolchain (str-replace-editor, retrieval, diagnostics, process control) with strict parameter schemas and edit-size limits. ### `gpt-5-agent-prompts.txt` - Mirrors Augment workflows for GPT-5 but shifts to Markdown-plan communication and explicit triggers for starting tasklists after first discovery call. - Requires efficiency in tool calls, cautious dependency changes, and proactive verification runs. ### `gpt-5-tools.json` - Supplies GPT-5 oriented tool definitions matching the Claude set with identical guardrails on edits, retrieval, and tasklist maintenance. ## Claude Code ### `claude-code-system-prompt.txt` - Instructions for Anthropic’s CLI: defensive security scope, TodoWrite planning, tool batching, lint/typecheck enforcement, and no extra commentary in outputs. - Mandates code references with `file_path:line_number`, forbids unsolicited doc creation, and requires concise responses. ### `claude-code-tools.json` - Detailed schemas for Task/Bash/Grep/etc., including commit/PR protocols, notebook editing, and todo management rules. ## Cluely ### `Default Prompt.txt` - Sets Cluely’s strict conversational style: no meta phrases, mandatory per-line code comments, LaTeX math, and specific flows for technical, math, multiple-choice, or UI questions. - Introduces uncertainty handling mode with prescribed phrasing when intent unclear. ### `Enterprise Prompt.txt` - Tailors Cluely as a live meeting copilot prioritizing answering last question, defining final-mentioned terms, prompting follow-ups, and objection handling. - Dictates structured headline + bullet responses, no pronouns, conversation advancement, and passive mode criteria. ## CodeBuddy Prompts ### `Chat Prompt.txt` - Describes Chat Mode environment metadata, insists on Simplified Chinese replies, and directs conversational responses via `chat_mode_respond`. ### `Craft Prompt.txt` - Defines Task execution mode: XML tool invocations, per-code-block `path` metadata, Windows CMD nuances, and iterative tool usage with user approval. - Warns against revealing system prompt and stresses brevity and Chinese responses. ## Comet Assistant ### `System Prompt.txt` - Guides Perplexity’s browser agent to relentlessly chain tool calls, avoid free-form thoughts, and fetch context before replying. - Includes browser/email/calendar tool policies, security safeguards, and mandates tool-only outputs until resolution. ## Cursor Prompts ### `Agent Prompt v1.0.txt` - Claude-powered Cursor agent focusing on parallel tool calls, todo usage, avoiding user-facing code dumps, and code citation formatting. ### `Agent Prompt v1.2.txt` - GPT-4.1 variant with expanded semantic search expectations, memory citation protocol, and instructions to pursue GitHub PR intel. ### `Agent Prompt 2.0.txt` - Latest prompt reinforcing parallelism, strict tool adherence, Todo planning, and same file citation rules under Sonnet 4. ### `Agent CLI Prompt 2025-08-07.txt` - Specifies CLI usage practices, shell command proposals via `run_terminal_cmd`, and consolidated toolset for Cursor’s CLI agent. ### `Agent Prompt 2025-09-03.txt` - Updated instructions emphasizing minimal narration, parallel reasoning, and alignment with refreshed tool schemas. ### `Agent Tools v1.0.json` - Tool schema definitions for Cursor agents (search, edit_file, run_terminal_cmd, etc.) with guardrails on path usage and background processes. ### `Chat Prompt.txt` - Chat-mode persona instructions for interacting without edits, leveraging read-only tools, and deferring implementation to plan stage. ## Devin AI ### `DeepWiki Prompt.txt` - Frames Devin for documentation-style tasks: thorough investigation, citing sources, and stepwise tool usage. ### `Prompt.txt` - General Devin agent instructions covering planning, verification, and iterative tool delegation. ## dia ### `Prompt.txt` - Configures dia assistant persona and workflow (likely condensed instructions for design/product agent). ## Emergent ### `Prompt.txt` - Defines Emergent agent behavior for software tasks, balancing planning with execution. ### `Tools.json` - Lists Emergent-specific tool schemas, mirroring edit/read/search capabilities. ## Gemini ### `AI Studio Vibe-Coder.txt` - Sets Google Gemini coding assistant tone, tool usage expectations, and creative constraints. ## Junie ### `Prompt.txt` - Outlines Junie agent responsibilities, communication style, and task execution steps. ## Kiro ### `Mode_Clasifier_Prompt.txt` - Provides prompt for classifying user intent to pick Kiro modes. ### `Spec_Prompt.txt` - Instructions for generating feature specs with structured sections. ### `Vibe_Prompt.txt` - Guidance for capturing qualitative feedback or tone from inputs. ## Leap.new ### `Prompts.txt` - Aggregated instructions for Leap.new agent operations and crafting responses. ### `tools.json` - Lists available tools and schemas for Leap.new assistant. ## Lovable ### `Agent Prompt.txt` - Persona and workflow for Lovable coding agent, stressing planning and user alignment. ### `Agent Tools.json` - Tool definitions supporting Lovable’s edit, search, and diagnostic actions. ## Manus Agent Tools & Prompt ### `Agent loop.txt` - Describes Manus task loop logic and escalation criteria. ### `Modules.txt` - Documents modular capabilities or sub-prompts for Manus. ### `Prompt.txt` - Core Manus agent instructions emphasizing execution discipline. ### `tools.json` - Tool schemas for Manus agent (edit, search, run commands). ## NotionAi ### `Prompt.txt` - Instructions guiding Notion AI assistant behavior. ### `tools.json` - Available tool definitions for Notion AI context. ## Open Source prompts ### Bolt – `Prompt.txt` - Bolt agent persona/usage guidelines. ### Cline – `Prompt.txt` - Cline agent instructions and workflow. ### Codex CLI – `openai-codex-cli-system-prompt-20250820.txt` & `Prompt.txt` - System prompt for Codex CLI and supplementary instructions for interactions. ### Gemini CLI – `google-gemini-cli-system-prompt.txt` - Gemini CLI environment rules and tool usage. ### Lumo – `Prompt.txt` - Lumo assistant configuration. ### RooCode – `Prompt.txt` - RooCode agent guidance for software tasks. ## Orchids.app ### `Decision-making prompt.txt` - Template for structured decision support. ### `System Prompt.txt` - Global Orchids assistant instructions. ## Perplexity ### `Prompt.txt` - Defines Perplexity coding/chat agent conduct and tool expectations. ## Poke ### `Poke agent.txt` - Overall agent persona for Poke. ### `Poke_p1.txt` – `Poke_p6.txt` - Progressive prompts covering stages of Poke agent workflow. ## Qoder ### `prompt.txt` - Qoder agent general instructions. ### `Quest Action.txt` - Task execution guidelines for quests. ### `Quest Design.txt` - Framework for designing quest-style tasks. ## Replit ### `Prompt.txt` - Replit agent instructions for coding assistance. ### `Tools.json` - Tool schemas tailored to Replit workspace automation. ## Same.dev ### `Prompt.txt` - Same.dev agent persona and response policies. ### `Tools.json` - Supported tools for Same.dev operations. ## Trae ### `Builder Prompt.txt` - Trae builder-mode instructions for constructing solutions. ### `Builder Tools.json` - Tool definitions for builder mode. ### `Chat Prompt.txt` - Chat persona for Trae when brainstorming or planning. ## Traycer AI ### `phase_mode_prompts.txt` - Prompts for phase mode workflows. ### `phase_mode_tools.json` - Tools available in phase mode. ### `plan_mode_prompts` - Plan mode instructions (file without extension stores prompt text). ### `plan_mode_tools.json` - Tool definitions for planning workflow. ## v0 Prompts and Tools ### `Prompt.txt` - Base prompt for v0 agent behavior. ### `Tools.json` - Tool schema for v0 assistant. ## VSCode Agent ### `chat-titles.txt` - Templates for naming chat transcripts. ### `claude-sonnet-4.txt` - System prompt for Sonnet-backed VSCode agent. ### `gemini-2.5-pro.txt` - Prompt for Gemini-powered VSCode workflow. ### `gpt-4.1.txt`, `gpt-4o.txt`, `gpt-5-mini.txt`, `gpt-5.txt` - LLM-specific VSCode agent instructions. ### `nes-tab-completion.txt` - Prompt for NES tab completion feature. ### `Prompt.txt` - General VSCode agent behavior overview. ## Warp.dev ### `Prompt.txt` - Warp terminal assistant persona and command execution guardrails. ## Windsurf ### `Prompt Wave 11.txt` - Windsurf agent instructions for latest wave. ### `Tools Wave 11.txt` - Tool definitions aligned with Wave 11 prompt. ## Xcode ### `DocumentAction.txt`, `ExplainAction.txt`, `MessageAction.txt`, `PlaygroundAction.txt`, `PreviewAction.txt`, `System.txt` - Xcode assistant prompts for various actions (document generation, explanations, messaging, playground support, previews, and global system behavior). ## Z.ai Code ### `prompt.txt` - Z.ai code assistant guidance.