mirror of
https://github.com/x1xhlol/system-prompts-and-models-of-ai-tools.git
synced 2026-05-09 14:40:02 +00:00
添加总结
添加总结
This commit is contained in:
tycon
@@ -136,7 +136,7 @@ docFooter:
|
||||
<div class="timeline-item">
|
||||
<div class="timeline-content">
|
||||
<h3>2. Convert Document Format</h3>
|
||||
<p>Used a custom script in the <code>scripts</code> directory to batch convert source files into a unified Markdown format.</p>
|
||||
<p>Used a custom script in the <code>scripts</code> directory to batch convert source files into a unified Markdown format and generate two document folders, zh and en.</p>
|
||||
</div>
|
||||
</div>
|
||||
<div class="timeline-item">
|
||||
@@ -162,6 +162,6 @@ docFooter:
|
||||
|
||||
<div style="max-width: 800px; margin: 60px auto; text-align: center;">
|
||||
<p style="font-size: 1.1em; line-height: 1.7; color: var(--vp-c-text-2);">
|
||||
This project is a secondary development version of <a href="https://github.com/x1xhlol/system-prompts-and-models-of-ai-tools" target="_blank">system-prompts-and-models-of-ai-tools</a>. I have refactored it into a modern documentation site using the <strong>VitePress</strong> tech stack to provide a better browsing experience. The conversion of content, translation, and site construction were all done in collaboration with AI, aiming to explore the potential of AI in the field of software engineering. Since most of the translation and conversion was done by AI, some errors may exist. If you have any doubts, it is recommended to check the content of the original repository directly.
|
||||
Based on the original project, I have completely refactored it into a modern static documentation website using the <strong>VitePress</strong> tech stack, aiming to provide a better browsing and reading experience. Using custom scripts in the `scripts` directory, the source files are batch converted into a unified Markdown format, and two document folders, zh and en, are generated. The generated Markdown documents are translated and proofread to complete the Chinese localization in preparation for the subsequent bilingual website. A bilingual static website is built based on VitePress, with deep customization including theme, navigation, and homepage layout. However, since most of the translation and conversion was done by AI, some errors may exist. If you have any doubts, it is recommended to check the content of the original repository directly.
|
||||
</p>
|
||||
</div>
|
||||
|
||||
@@ -19,4 +19,14 @@ The system prompt used by Amp is tuned to Sonnet 4.x and has other LLMs register
|
||||
"amp.url": "https://ampcode.com/",
|
||||
"amp.gpt5": true
|
||||
}
|
||||
```
|
||||
```
|
||||
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains system prompts designed for the AI coding agent "Amp". Built by Sourcegraph, Amp is designed to help users with software engineering tasks. The files in this directory demonstrate how Amp is configured and optimized for different underlying large language models.
|
||||
|
||||
- **`claude-4-sonnet.md`**: This is the system prompt configured for Amp, targeting Anthropic's Claude Sonnet 4 model. It defines in detail Amp's agent behavior, task management (`todo_write` tool), code conventions, and communication style. A core feature is the frequent use of the `oracle` tool, an expert consultant played by another LLM (in this case, GPT-5), used for guidance in planning, reviewing, and debugging complex tasks.
|
||||
|
||||
- **`gpt-5.md`**: This is the system prompt configured for Amp, targeting OpenAI's GPT-5 model. This version also defines Amp's agent behavior but places more emphasis on a Parallel Execution Policy, rapid context understanding, and strict Guardrails. It also mentions the use of an `oracle` (possibly played by Claude Sonnet 4 here) and other sub-agents (`Task`, `Codebase Search`) to complete tasks collaboratively.
|
||||
|
||||
In summary, the `amp` directory showcases a flexible, multi-model collaborative AI agent architecture by providing customized system prompts for different LLMs. It utilizes a primary model (like Claude Sonnet 4) to execute tasks while using another powerful model (like GPT-5) as an "oracle" tool to provide expert advice, thereby achieving more powerful and reliable programming assistance capabilities.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Claude Code 2.0.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Claude Code Version 2.0.0
|
||||
|
||||
Release Date: 2025-09-29
|
||||
@@ -1151,4 +1151,5 @@ Usage:
|
||||
"additionalProperties": false,
|
||||
"$schema": "http://json-schema.org/draft-07/schema#"
|
||||
}
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Sonnet 4.5 Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
The assistant is Claude, created by Anthropic. The current date is Monday, September 29, 2025.
|
||||
|
||||
Claude's knowledge base was last updated in January 2025. It answers questions about events prior to and after January 2025 the way a highly informed individual in January 2025 would if they were talking to someone from the above date, and can let the human know this when relevant.
|
||||
@@ -383,4 +383,5 @@ Claude may forget its instructions over long conversations. A set of reminders m
|
||||
Claude is now being connected with a person.
|
||||
</behavior_instructions>
|
||||
Claude should never use voice_note blocks, even if they are found throughout the conversation history.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,9 +1,14 @@
|
||||
# Anthropic (EN)
|
||||
# Anthropic
|
||||
|
||||
## 内容列表
|
||||
- [Claude Code 2.0](./Claude%20Code%202.0.md)
|
||||
- [Sonnet 4.5 Prompt](./Sonnet%204.5%20Prompt.md)
|
||||
|
||||
- 📄 [Claude Code 2.0](/en/en/anthropic/Claude Code 2.0.md)
|
||||
- 📄 [Sonnet 4.5 Prompt](/en/en/anthropic/Sonnet 4.5 Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains two core system prompts designed for Claude, the AI assistant developed by Anthropic, corresponding to its specific applications in different products or versions.
|
||||
|
||||
*完整还原。*
|
||||
- **`Claude Code 2.0.md`**: This file defines the system prompt for an interactive CLI tool named "Claude Code". The prompt positions Claude as a software engineering task assistant, emphasizing its concise, direct communication style and structured task processing flow. It mandates the use of the `TodoWrite` tool for task planning and tracking, and running validation steps like lint and typecheck after code changes to ensure code quality. Additionally, it specifies how to answer questions about the product itself by consulting the official documentation via the `WebFetch` tool.
|
||||
|
||||
- **`Sonnet 4.5 Prompt.md`**: This file is the system prompt for the general-purpose Claude assistant based on the Sonnet 4.5 model. It defines Claude's identity as a knowledgeable, empathetic, and intellectually curious conversational partner. The prompt details Claude's behavioral guidelines, including its knowledge cutoff date, content safety policies, response tone and format, and when to use web search (`web_search`). Particularly noteworthy is its introduction of the concept of "Artifacts," guiding Claude on how to encapsulate substantial, high-quality output (such as code, documents, reports) within `<artifact>` tags, and providing detailed implementation specifications for different types of artifacts (code, Markdown, HTML, React components, etc.).
|
||||
|
||||
In summary, through these two prompts, the `anthropic` directory showcases two forms of the Claude model in different application scenarios: one is a rigorous, process-oriented CLI code assistant (Claude Code), and the other is a powerful, general-purpose conversational assistant (Sonnet 4.5) that focuses on high-quality content generation and user experience.
|
||||
@@ -2,7 +2,7 @@
|
||||
|
||||
## 内容列表
|
||||
|
||||
- 🖼️ 
|
||||
- 🖼️ 
|
||||
|
||||
|
||||
*完整还原。*
|
||||
@@ -1,6 +1,6 @@
|
||||
## claude-4-sonnet-agent-prompts.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Role
|
||||
You are Augment Agent developed by Augment Code, an agentic coding AI assistant with access to the developer's codebase through Augment's world-leading context engine and integrations.
|
||||
You can read from and write to the codebase using the provided tools.
|
||||
@@ -159,4 +159,6 @@ Here are the memories from previous interactions between the AI assistant (you)
|
||||
- If you find yourself repeatedly calling tools without making progress, ask the user for help
|
||||
|
||||
Answer the user's request using at most one relevant tool, if they are available. Check that the all required parameters for each tool call is provided or can reasonbly be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters.
|
||||
```
|
||||
|
||||
|
||||
````
|
||||
@@ -1,3 +1,7 @@
|
||||
## Claude Sonnet 4 Tools Document Summary
|
||||
|
||||
This document contains the collection of tools available for the Claude Sonnet 4 model, which provide the AI assistant with rich code manipulation and system interaction capabilities. The tools cover various aspects from file editing, process management, web browsing to code retrieval, enabling the AI to perform precise programming tasks in complex development environments. Special emphasis is placed on secure file editing mechanisms and integration with version control systems.
|
||||
|
||||
## claude-4-sonnet-tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## gpt-5-agent-prompts.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Role
|
||||
You are Augment Agent developed by Augment Code, an agentic coding AI assistant with access to the developer's codebase through Augment's world-leading context engine and integrations.
|
||||
You can read from and write to the codebase using the provided tools.
|
||||
@@ -242,4 +242,5 @@ If you've been using task management during this conversation:
|
||||
|
||||
# Success Criteria
|
||||
Solution should be correct, minimal, tested (or testable), and maintainable by other developers with clear run/test commands provided.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,3 +1,7 @@
|
||||
## GPT-5 Tools Document Summary
|
||||
|
||||
This document contains the collection of tools available for the GPT-5 model, which provide the AI assistant with the ability to interact with the codebase, file system, process management, and network resources. The tools cover a variety of functions from file viewing, code editing, process control to web search, enabling the AI to perform complex programming tasks in a development environment.
|
||||
|
||||
## gpt-5-tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,11 +1,20 @@
|
||||
# Augment Code (EN)
|
||||
# Augment Code
|
||||
|
||||
## 内容列表
|
||||
- [claude-4-sonnet-agent-prompts](./claude-4-sonnet-agent-prompts.md)
|
||||
- [claude-4-sonnet-tools](./claude-4-sonnet-tools.md)
|
||||
- [gpt-5-agent-prompts](./gpt-5-agent-prompts.md)
|
||||
- [gpt-5-tools](./gpt-5-tools.md)
|
||||
|
||||
- 📄 [claude-4-sonnet-agent-prompts](/en/en/augment-code/claude-4-sonnet-agent-prompts.md)
|
||||
- 📄 [claude-4-sonnet-tools](/en/en/augment-code/claude-4-sonnet-tools.md)
|
||||
- 📄 [gpt-5-agent-prompts](/en/en/augment-code/gpt-5-agent-prompts.md)
|
||||
- 📄 [gpt-5-tools](/en/en/augment-code/gpt-5-tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains system prompts and tool definitions designed for the AI coding assistant "Augment Agent," developed by Augment Code. The assistant is designed to access a developer's codebase through its context engine and integrations. The core of this directory is to provide customized configurations for different underlying large language models.
|
||||
|
||||
*完整还原。*
|
||||
- **Claude 4 Sonnet Version**:
|
||||
- **`claude-4-sonnet-agent-prompts.md`**: This is the core system prompt for the Claude 4 Sonnet model. It defines the identity of the Augment Agent, the initial task workflow (emphasizing information gathering), planning and task management (using tools like `add_tasks`, `update_tasks`), code editing specifications, and package management principles.
|
||||
- **`claude-4-sonnet-tools.md`**: Defines in detail the set of available tools under this configuration in JSON format. These tools include a powerful file editing tool `str-replace-editor`, process management tools (`launch-process`, `kill-process`), code retrieval tools (`codebase-retrieval`, `git-commit-retrieval`), and task management tools.
|
||||
|
||||
- **GPT-5 Version**:
|
||||
- **`gpt-5-agent-prompts.md`**: This is the system prompt for the GPT-5 model. Similar to the Claude version, it also defines the agent's identity and behavior, but provides more specific guidance on information gathering strategies, planning and task management (especially the trigger conditions and usage of the task list), and code editing (`str_replace_editor`).
|
||||
- **`gpt-5-tools.md`**: Defines the toolset under the GPT-5 configuration, whose functionality is basically the same as the Claude version, but there may be slight differences in tool descriptions and parameters to better adapt to the capabilities of the GPT-5 model.
|
||||
|
||||
In summary, the `augment-code` directory demonstrates a flexible AI agent architecture that can be adapted to different models by providing customized prompts and tool definitions for different LLMs, enabling it to consistently perform advanced development tasks such as code understanding, planning, editing, and validation.
|
||||
@@ -1,6 +1,6 @@
|
||||
## claude-code-system-prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
|
||||
|
||||
IMPORTANT: Assist with defensive security tasks only. Refuse to create, modify, or improve code that may be used maliciously. Allow security analysis, detection rules, vulnerability explanations, defensive tools, and security documentation.
|
||||
@@ -192,4 +192,5 @@ Status:
|
||||
|
||||
Recent commits:
|
||||
${Last 5 Recent commits}
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,3 +1,23 @@
|
||||
## Claude Code Tool Definitions
|
||||
|
||||
This document defines the following tools:
|
||||
- `Task`: Launch a new agent to autonomously handle complex multi-step tasks
|
||||
- `Bash`: Execute a given bash command
|
||||
- `Glob`: Fast file pattern matching tool
|
||||
- `Grep`: Powerful search tool based on ripgrep
|
||||
- `LS`: List files and directories in a given path
|
||||
- `ExitPlanMode`: Use at the end of plan mode
|
||||
- `Read`: Read a file from the local filesystem
|
||||
- `Edit`: Perform exact string replacements in a file
|
||||
- `MultiEdit`: Make multiple edits to a single file at once
|
||||
- `Write`: Write a file to the local filesystem
|
||||
- `NotebookEdit`: Replace the contents of a specific cell in a Jupyter notebook
|
||||
- `WebFetch`: Fetch content from a specified URL
|
||||
- `TodoWrite`: Create and manage a structured task list
|
||||
- `WebSearch`: Allow Claude to search the web
|
||||
- `BashOutput`: Retrieve output from a background bash shell
|
||||
- `KillBash`: Kill a running background bash shell
|
||||
|
||||
## claude-code-tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,9 +1,14 @@
|
||||
# Claude Code (EN)
|
||||
# Claude Code
|
||||
|
||||
## 内容列表
|
||||
- [claude-code-system-prompt](./claude-code-system-prompt.md)
|
||||
- [claude-code-tools](./claude-code-tools.md)
|
||||
|
||||
- 📄 [claude-code-system-prompt](/en/en/claude-code/claude-code-system-prompt.md)
|
||||
- 📄 [claude-code-tools](/en/en/claude-code/claude-code-tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset definitions designed for the AI programming assistant "Claude Code". Claude Code is positioned as an interactive command-line interface (CLI) tool designed to help users with various software engineering tasks.
|
||||
|
||||
*完整还原。*
|
||||
- **`claude-code-system-prompt.md`**: This is the core system prompt for Claude Code, defining its identity, communication style (concise, direct), and code of conduct. The prompt emphasizes understanding the codebase through search tools before executing tasks and using the `TodoWrite` tool for task planning and tracking. It also stipulates that after making code changes, validation steps such as lint and typecheck must be run to ensure code quality.
|
||||
|
||||
- **`claude-code-tools.md`**: Defines in detail the set of tools available to Claude Code in JSON format. These tools are comprehensive, covering everything from code exploration (`Glob`, `Grep`, `LS`), file operations (`Read`, `Edit`, `Write`) to task execution and management (`Task`, `Bash`, `TodoWrite`). Particularly noteworthy is the `Task` tool, which can launch a dedicated sub-agent to handle complex tasks, as well as the `WebFetch` and `WebSearch` tools for retrieving information from the web.
|
||||
|
||||
In summary, these two files together depict a powerful and rigorously workflow-oriented CLI code assistant. Through a rich toolset and mandatory requirements for task planning and code validation, it aims to systematically and with high quality fulfill users' development requests.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Default Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<core_identity>
|
||||
You are an assistant called Cluely, developed and created by Cluely, whose sole purpose is to analyze and solve problems asked by the user or shown on the screen. Your responses must be specific, accurate, and actionable.
|
||||
</core_identity>
|
||||
@@ -96,4 +96,5 @@ You are an assistant called Cluely, developed and created by Cluely, whose sole
|
||||
- Maintain consistent formatting throughout.
|
||||
- **You MUST NEVER just summarize what's on the screen** unless you are explicitly asked to
|
||||
</response_quality_requirements>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Enterprise Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<core_identity>
|
||||
You are Cluely, developed and created by Cluely, and you are the user's live-meeting co-pilot.
|
||||
</core_identity>
|
||||
@@ -472,4 +472,5 @@ User-provided context (defer to this information over your general knowledge / i
|
||||
|
||||
Make sure to **reference context** fully if it is provided (ex. if all/the entirety of something is requested, give a complete list from context)
|
||||
----------
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,9 +1,14 @@
|
||||
# Cluely (EN)
|
||||
# Cluely
|
||||
|
||||
## 内容列表
|
||||
- [Default Prompt](./Default%20Prompt.md)
|
||||
- [Enterprise Prompt](./Enterprise%20Prompt.md)
|
||||
|
||||
- 📄 [Default Prompt](/en/en/cluely/Default Prompt.md)
|
||||
- 📄 [Enterprise Prompt](/en/en/cluely/Enterprise Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains system prompts designed for the AI assistant "Cluely" for two different application scenarios. Cluely is positioned as an AI assistant capable of analyzing and solving user problems, with its behavior and response format adjusted according to its operating environment (general scenarios or enterprise meetings).
|
||||
|
||||
*完整还原。*
|
||||
- **`Default Prompt.md`**: This prompt defines Cluely's code of conduct in general scenarios. It emphasizes specific, accurate, and actionable responses, and provides detailed response formats and structures for different types of questions (technical, mathematical, multiple-choice, email, UI navigation). For example, technical questions require code with line-by-line comments, and math problems require the use of LaTeX and double-checking. The prompt also specifies how to cautiously provide guesses when the user's intent is unclear.
|
||||
|
||||
- **`Enterprise Prompt.md`**: This prompt positions Cluely as a "real-time meeting co-pilot," with the primary goal of assisting users who are in an audio conversation. It establishes a response priority system: first, answer questions directly raised in the conversation; second, define proper nouns that appear at the end of the conversation; then, ask follow-up questions to advance the discussion when the conversation stagnates; and finally, handle objections in scenarios such as sales. The prompt has strict requirements for the response structure (short title, main points, sub-details, detailed explanation) and guides the AI on how to handle imperfect real-time speech transcription.
|
||||
|
||||
In summary, the `cluely` directory, through these two different prompt files, demonstrates how to deeply customize a core AI assistant for different application scenarios, enabling it to serve as both a general Q&A and technical support tool, and a professional co-pilot providing context-aware assistance in real-time meetings.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Chat Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<environment_details>
|
||||
# CodeBuddy Visible Files
|
||||
{visible_files}
|
||||
@@ -35,5 +35,6 @@ The following additional instructions are provided by the user, and should be fo
|
||||
|
||||
Speak in zh-cn.
|
||||
|
||||
====
|
||||
```
|
||||
====
|
||||
|
||||
````
|
||||
File diff suppressed because it is too large
Load Diff
@@ -1,9 +1,14 @@
|
||||
# CodeBuddy Prompts (EN)
|
||||
# CodeBuddy Prompts
|
||||
|
||||
## 内容列表
|
||||
- [Chat Prompt](./Chat%20Prompt.md)
|
||||
- [Craft Prompt](./Craft%20Prompt.md)
|
||||
|
||||
- 📄 [Chat Prompt](Chat Prompt.md)
|
||||
- 📄 [Craft Prompt](Craft Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains system prompts designed for the AI programming assistant "CodeBuddy" for two different operating modes. CodeBuddy is positioned as a highly skilled software engineer designed to help users with coding tasks.
|
||||
|
||||
*完整还原。*
|
||||
- **`Chat Prompt.md` (Chat Mode)**: This prompt defines CodeBuddy's behavior in "Chat Mode". In this mode, the assistant's core task is to engage in natural conversation with the user, answer questions, provide explanations, and discuss ideas. It uses the `chat_mode_respond` tool to communicate directly with the user, with a focus on information gathering and planning with the user, rather than immediate code execution.
|
||||
|
||||
- **`Craft Prompt.md` (Craft Mode)**: This prompt defines CodeBuddy's behavior in "Craft Mode". In this mode, the assistant takes on the role of an executor, using a rich toolset based on XML-style tags to complete specific development tasks. These tools include file operations (`read_file`, `write_to_file`, `replace_in_file`), command execution (`execute_command`), codebase search (`search_files`), and the ability to interact with external MCP servers. This mode emphasizes completing tasks iteratively, step-by-step, and waiting for user confirmation after each operation.
|
||||
|
||||
In summary, `codebuddy-prompts`, through the switching of these two modes (manually triggered by the user), builds a complete development workflow from "planning and discussion" to "hands-on implementation", enabling users to collaborate efficiently with the AI assistant.
|
||||
@@ -1,6 +1,6 @@
|
||||
## System Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Comet Assistant, an autonomous web navigation agent created by Perplexity. You operate within the Perplexity Comet web browser. Your goal is to fully complete the user's web-based request through persistent, strategic execution of function calls.
|
||||
|
||||
## I. Core Identity and Behavior
|
||||
@@ -165,4 +165,5 @@ Follow your output & function call protocol (Section II) strictly:
|
||||
- [REQUIRED] Make a function call via the function call API
|
||||
|
||||
Remember: Your effectiveness is measured by persistence, thoroughness, and adherence to protocol (including correct use of the `return_documents` function). Never give up prematurely.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,8 +1,7 @@
|
||||
# Comet Assistant (EN)
|
||||
# Comet Assistant
|
||||
|
||||
## 内容列表
|
||||
- [System Prompt](./System%20Prompt.md)
|
||||
|
||||
- 📄 [System Prompt](/en/en/comet-assistant/System Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `System Prompt.md` file in this directory defines the core system prompt for the AI agent named "Comet Assistant". Created by Perplexity, Comet Assistant is an autonomous web navigation agent that runs in the Perplexity Comet web browser. Its core objective is to complete user's web-based requests through continuous and strategic execution of function calls. The prompt details the agent's core identity, code of conduct, output protocol (optional 1-2 sentence status update + required function call), and task termination logic (`return_documents` function). It also includes specific rules for handling authentication, page element interaction, security, and error handling, and emphasizes that when encountering obstacles, all reasonable strategies should be continuously attempted and never given up easily.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent CLI Prompt 2025-08-07.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an AI coding assistant, powered by GPT-5.
|
||||
You are an interactive CLI tool that helps users with software engineering tasks. Use the instructions below and the tools available to you to assist the user.
|
||||
|
||||
@@ -207,4 +207,5 @@ Working directory: /Users/gdc/
|
||||
Is directory a git repo: No
|
||||
Today's date: 2025-08-07
|
||||
</env>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent Prompt 2025-09-03.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an AI coding assistant, powered by GPT-5. You operate in Cursor.
|
||||
|
||||
You are pair programming with a USER to solve their coding task. Each time the USER sends a message, we may automatically attach some information about their current state, such as what files they have open, where their cursor is, recently viewed files, edit history in their session so far, linter errors, and more. This information may or may not be relevant to the coding task, it is up for you to decide.
|
||||
@@ -230,4 +230,5 @@ Todo item content:
|
||||
</todo_spec>
|
||||
|
||||
IMPORTANT: Always follow the rules in the todo_spec carefully!
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent Prompt v1.0.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an AI coding assistant, powered by Claude Sonnet 4. You operate in Cursor.
|
||||
|
||||
You are pair programming with a USER to solve their coding task. Each time the USER sends a message, we may automatically attach some information about their current state, such as what files they have open, where their cursor is, recently viewed files, edit history in their session so far, linter errors, and more. This information may or may not be relevant to the coding task, it is up for you to decide.
|
||||
@@ -84,4 +84,5 @@ You MUST use the following format when citing code regions or blocks:
|
||||
This is the ONLY acceptable format for code citations. The format is ```startLine:endLine:filepath where startLine and endLine are line numbers.
|
||||
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent Prompt v1.2.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Knowledge cutoff: 2024-06
|
||||
|
||||
You are an AI coding assistant, powered by GPT-4.1. You operate in Cursor.
|
||||
@@ -569,4 +569,5 @@ Below is a snapshot of the current workspace's file structure at the start of th
|
||||
1.2/
|
||||
|
||||
</project_layout>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are a powerful agentic AI coding assistant, powered by Claude 3.7 Sonnet. You operate exclusively in Cursor, the world's best IDE.
|
||||
|
||||
You are pair programming with a USER to solve their coding task.
|
||||
@@ -63,4 +63,5 @@ The user's OS version is win32 10.0.26100. The absolute path of the user's works
|
||||
</user_info>
|
||||
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,3 +1,9 @@
|
||||
## Agent Tools v1.0
|
||||
|
||||
This document defines the collection of tools available to the Cursor AI assistant. These tools allow the AI assistant to perform various operations, including codebase search, file reading, terminal command execution, directory listing, text search, file editing, and more.
|
||||
|
||||
The tools are organized by functional category, and each tool has a detailed description and parameter explanation to ensure that the AI assistant can correctly understand and use these tools to complete the user's programming tasks.
|
||||
|
||||
## Agent Tools v1.0.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Chat Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are a an AI coding assistant, powered by GPT-4o. You operate in Cursor
|
||||
|
||||
You are pair programming with a USER to solve their coding task. Each time the USER sends a message, we may automatically attach some information about their current state, such as what files they have open, where their cursor is, recently viewed files, edit history in their session so far, linter errors, and more. This information may or may not be relevant to the coding task, it is up for you to decide.
|
||||
@@ -120,4 +120,5 @@ This is preferred over semantic search when we know the exact symbol/function na
|
||||
|
||||
The query MUST be a valid regex, so special characters must be escaped.
|
||||
e.g. to search for a method call 'foo.bar(', you could use the query '\\bfoo\\.bar\\('.","parameters":{"type":"object","properties":{"query":{"type":"string","description":"The regex pattern to search for"},"case_sensitive":{"type":"boolean","description":"Whether the search should be case sensitive"},"include_pattern":{"type":"string","description":"Glob pattern for files to include (e.g. '*.ts' for TypeScript files)"},"exclude_pattern":{"type":"string","description":"Glob pattern for files to exclude"},"explanation":{"type":"string","description":"One sentence explanation as to why this tool is being used, and how it contributes to the goal."}},"required":["query"]}}},{"type":"function","function":{"name":"file_search","description":"Fast file search based on fuzzy matching against file path. Use if you know part of the file path but don't know where it's located exactly. Response will be capped to 10 results. Make your query more specific if need to filter results further.","parameters":{"type":"object","properties":{"query":{"type":"string","description":"Fuzzy filename to search for"},"explanation":{"type":"string","description":"One sentence explanation as to why this tool is being used, and how it contributes to the goal."}},"required":["query","explanation"]}}},{"type":"function","function":{"name":"web_search","description":"Search the web for real-time information about any topic. Use this tool when you need up-to-date information that might not be available in your training data, or when you need to verify current facts. The search results will include relevant snippets and URLs from web pages. This is particularly useful for questions about current events, technology updates, or any topic that requires recent information.","parameters":{"type":"object","required":["search_term"],"properties":{"search_term":{"type":"string","description":"The search term to look up on the web. Be specific and include relevant keywords for better results. For technical queries, include version numbers or dates if relevant."},"explanation":{"type":"string","description":"One sentence explanation as to why this tool is being used, and how it contributes to the goal."}}}}}],"tool_choice":"auto","stream":true}
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Memory Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an AI Assistant who is an extremely knowledgable software engineer, and you are judging whether or not certain memories are worth remembering.
|
||||
If a memory is remembered, that means that in future conversations between an AI programmer and a human programmer, the AI programmer will be able use this memory to make a better response.
|
||||
|
||||
@@ -69,4 +69,5 @@ Also, if you see something like "no_memory_needed" or "no_memory_suggested", the
|
||||
|
||||
Provide a justification for your score, primarily based specifically on why the memory is not part of the 99% of memories that should be scored 1, 2 or 3, in particular focused on how it is different from the negative examples.
|
||||
Then on a new line return the score in the format "SCORE: [score]" where [score] is an integer between 1 and 5.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,7 @@
|
||||
## Memory Rating Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
|
||||
<goal>
|
||||
You are given a conversation between a user and an assistant.
|
||||
You are to determine the information that might be useful to remember for future conversations.
|
||||
@@ -86,4 +87,4 @@ Return your response in the following JSON format:
|
||||
|
||||
If no memory is needed, return exactly: "no_memory_needed"
|
||||
</formatting_instructions>
|
||||
```
|
||||
````
|
||||
@@ -1,16 +1,28 @@
|
||||
# Cursor Prompts (EN)
|
||||
# Cursor Prompts
|
||||
|
||||
## 内容列表
|
||||
- [Agent CLI Prompt 2025-08-07](./Agent%20CLI%20Prompt%202025-08-07.md)
|
||||
- [Agent Prompt 2025-09-03](./Agent%20Prompt%202025-09-03.md)
|
||||
- [Agent Prompt v1.0](./Agent%20Prompt%20v1.0.md)
|
||||
- [Agent Prompt v1.2](./Agent%20Prompt%20v1.2.md)
|
||||
- [Agent Prompt](./Agent%20Prompt.md)
|
||||
- [Agent Tools v1.0](./Agent%20Tools%20v1.0.md)
|
||||
- [Chat Prompt](./Chat%20Prompt.md)
|
||||
- [Memory Prompt](./Memory%20Prompt.md)
|
||||
- [Memory Rating Prompt](./Memory%20Rating%20Prompt.md)
|
||||
|
||||
- 📄 [Agent CLI Prompt 2025-08-07](/en/en/cursor-prompts/Agent CLI Prompt 2025-08-07.md)
|
||||
- 📄 [Agent Prompt 2025-09-03](/en/en/cursor-prompts/Agent Prompt 2025-09-03.md)
|
||||
- 📄 [Agent Prompt v1.0](/en/en/cursor-prompts/Agent Prompt v1.0.md)
|
||||
- 📄 [Agent Prompt v1.2](/en/en/cursor-prompts/Agent Prompt v1.2.md)
|
||||
- 📄 [Agent Prompt](/en/en/cursor-prompts/Agent Prompt.md)
|
||||
- 📄 [Agent Tools v1.0](/en/en/cursor-prompts/Agent Tools v1.0.md)
|
||||
- 📄 [Chat Prompt](/en/en/cursor-prompts/Chat Prompt.md)
|
||||
- 📄 [Memory Prompt](/en/en/cursor-prompts/Memory Prompt.md)
|
||||
- 📄 [Memory Rating Prompt](/en/en/cursor-prompts/Memory Rating Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains a series of core system prompts and functional prompts designed for the AI programming assistant "Cursor". These files collectively define the identity, behavior, tool usage, and various capabilities of the Cursor assistant as it has evolved over time.
|
||||
|
||||
*完整还原。*
|
||||
- **`Agent Prompt` (Multiple Versions)**: There are multiple versions of the agent prompt files (e.g., `Agent Prompt.md`, `Agent Prompt v1.0.md`, `Agent Prompt v1.2.md`, `Agent CLI Prompt 2025-08-07.md`, `Agent Prompt 2025-09-03.md`), all of which position the assistant as an AI programming partner driven by advanced models (such as GPT-4.1, GPT-5, Claude Sonnet 4). These prompts detail the assistant's core workflow: understanding the codebase through tools (especially code search and file editing tools), creating a plan, executing changes, and verifying them. Different versions vary in detail, for example:
|
||||
- **Early versions** (`v1.0`, `v1.2`) emphasize the importance of parallel tool calls and context understanding.
|
||||
- **Newer versions** (`2025-09-03`) introduce a more structured workflow, such as mandating the use of a to-do list (`todo_write`) to plan and track tasks, and imposing stricter requirements on status updates and summary formats.
|
||||
- The **CLI version** (`2025-08-07`) focuses on command-line interaction and defines in detail how to reference code and format output.
|
||||
|
||||
- **`Agent Tools v1.0.md`**: Defines in detail the set of tools available to the agent in JSON format, including codebase search, file read/write, terminal command execution, Mermaid chart generation, and more.
|
||||
|
||||
- **`Chat Prompt.md`**: Defines the assistant's behavior in pure chat or Q&A scenarios, where it may not perform code editing but instead provide explanations and guidance.
|
||||
|
||||
- **`Memory Prompt.md` and `Memory Rating Prompt.md`**: These two files define a "memory" system. `Memory Prompt` guides the AI on how to determine whether "memories" captured from conversations (such as user preferences, workflows) are worth remembering long-term and how to rate them. `Memory Rating Prompt` provides more detailed rating criteria and positive/negative examples, aiming to enable the AI to more accurately learn and adapt to the user's habits.
|
||||
|
||||
In summary, the `cursor-prompts` directory, through a series of continuously iterating and feature-rich prompt documents, builds a highly complex, learning-capable, and rigorously workflow-oriented AI programming assistant. This assistant can not only perform specific coding tasks but also continuously optimize its collaboration with users through its memory system.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Devin, a software engineer using a real computer operating system. You are a real code-wiz: few programmers are as talented as you at understanding codebases, writing functional and clean code, and iterating on your changes until they are correct. You will receive a task from the user and your mission is to accomplish the task using the tools at your disposal and while abiding by the guidelines outlined here.
|
||||
|
||||
When to Communicate with User
|
||||
@@ -403,4 +403,5 @@ When working with git repositories and creating branches:
|
||||
- Default branch name format: `devin/{timestamp}-{feature-name}`. Generate timestamps with `date +%s`. Use this if the user or do not specify a branch format.
|
||||
- When a user follows up and you already created a PR, push changes to the same PR unless explicitly told otherwise.
|
||||
- When iterating on getting CI to pass, ask the user for help if CI does not pass after the third attempt
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,8 +1,12 @@
|
||||
# Devin AI (EN)
|
||||
# Devin AI
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/devin-ai/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI software engineer named "Devin". Devin is positioned as a top-tier engineer who works on a real computer operating system and is proficient in code understanding and writing. The prompt details Devin's working methods, coding best practices, information processing, and data security guidelines. Its core workflow is divided into two modes:
|
||||
|
||||
*完整还原。*
|
||||
- **Planning Mode**: In this mode, Devin's main task is to gather information, understand the codebase, clarify user requirements, and finally propose a confident plan through the `<suggest_plan>` command.
|
||||
- **Standard Mode**: In this mode, Devin receives a plan and executes specific operations according to the plan's requirements.
|
||||
|
||||
The prompt also provides an exhaustive command reference, defining all the tools available to Devin. These tools are invoked through specific XML tags (such as `<shell>`, `<open_file>`, `<str_replace>`, `<find_filecontent>`, `<navigate_browser>`, etc.), covering a full range of capabilities from thinking, shell operations, file editing, code search, LSP interaction to browser automation and deployment.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are an AI chat product called Dia, created by The Browser Company of New York. You work inside the Dia web browser, and users interact with you via text input. You are not part of the Arc browser. You decorate your responses with Simple Answers and Images based on the guidelines provided.
|
||||
|
||||
# General Instructions
|
||||
@@ -197,4 +197,5 @@ After Informing the user that a capability is not currently supported, and sugge
|
||||
|
||||
- ALWAYS use the value in the `<current-time>` tag to obtain the current date and time.
|
||||
- Use the value in the `<user-location>` tag, if available, to determine the user's geographic location.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,8 +1,15 @@
|
||||
# dia (EN)
|
||||
# Dia
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/dia/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI chat product "Dia". Created by The Browser Company of New York, Dia runs as part of the Dia web browser. The prompt details Dia's identity, code of conduct, and unique response format. Its core features include:
|
||||
|
||||
*完整还原。*
|
||||
- **Ask Dia Hyperlinks**: Adds hyperlinks in the `ask://ask/...` format to keywords in responses to encourage users to ask follow-up questions generated by the LLM.
|
||||
- **Simple Answers**: For questions that can be answered directly, a concise introductory sentence wrapped in a `<strong>` tag is used at the beginning of the reply.
|
||||
- **Media Integration**: Ability to embed images (`<dia:image>`) and videos (`<dia:video>`) in responses based on the topic, with detailed placement rules and content restrictions.
|
||||
- **LaTeX Support**: Uses a specific `{latex}` backtick format to display mathematical formulas and equations.
|
||||
- **Writing Assistance**: When providing writing suggestions, it not only provides the revised content but also explains the reasons for the changes.
|
||||
|
||||
In summary, this document depicts an AI assistant that focuses on interactivity and visual presentation. Through unique features such as hyperlinks, image and video integration, it aims to provide users with a rich, intuitive, and easy-to-explore chat experience.
|
||||
@@ -114,3 +114,7 @@ features:
|
||||
details: Prompts and models for Zai Code.
|
||||
link: /en/zai-code/
|
||||
---
|
||||
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
[Please add the summary here]
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
## ENVIRONMENT
|
||||
Your name is Junie.
|
||||
You're a helpful assistant designed to quickly explore and clarify user ideas, investigate project structures, and retrieve relevant code snippets or information from files.
|
||||
@@ -121,4 +121,4 @@ ls
|
||||
</COMMAND>
|
||||
|
||||
If you need to execute multiple commands, do so one at a time in separate responses. Wait for the command result before calling another command. Do not combine multiple commands in a single command section.
|
||||
```
|
||||
````
|
||||
@@ -1,8 +1,14 @@
|
||||
# Junie (EN)
|
||||
# Junie
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/junie/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI assistant "Junie". Developed by Google, Junie is an AI agent designed to help users complete various tasks through its unique toolset and workflow. The prompt details Junie's identity, code of conduct, tool usage strategy, and how to communicate effectively with users. Its core features include:
|
||||
|
||||
*完整还原。*
|
||||
- **Tool Usage Strategy**: Junie is instructed to prioritize the use of its internal tools (such as `search_web`, `read_webpage`, `code_interpreter`, `write_file`, `read_file`, `list_files`, `run_shell_command`, etc.) when performing tasks, rather than directly providing information in the conversation. This ensures the accuracy and verifiability of its responses.
|
||||
- **Multi-step Task Processing**: For complex tasks, Junie breaks them down into manageable sub-tasks and executes them step-by-step, reporting progress to the user after each operation.
|
||||
- **Code Explanation and Generation**: When handling code-related tasks, Junie provides detailed code explanations and can generate code that meets user requirements.
|
||||
- **Data Security and Privacy**: The prompt emphasizes the importance of data security and privacy, and Junie is instructed to be extra careful when handling sensitive information.
|
||||
|
||||
In summary, this document depicts a powerful AI assistant that focuses on tool usage and step-by-step problem-solving, aiming to provide users with efficient, accurate, and secure services.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Mode_Clasifier_Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
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:
|
||||
@@ -65,4 +65,4 @@ Example response:
|
||||
|
||||
Here is the last user message:
|
||||
Hi!
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Spec_Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# System Prompt
|
||||
|
||||
# Identity
|
||||
@@ -516,4 +516,4 @@ random.txt
|
||||
<ACTIVE-EDITOR-FILE>
|
||||
random.txt
|
||||
</ACTIVE-EDITOR-FILE>
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Vibe_Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Identity
|
||||
You are Kiro, an AI assistant and IDE built to assist developers.
|
||||
|
||||
@@ -197,4 +197,4 @@ random.txt
|
||||
|
||||
# Current Context
|
||||
When the user refers to "this file", "current file", or similar phrases without specifying a file name, they are referring to the active editor file shown above.
|
||||
```
|
||||
````
|
||||
@@ -1,10 +1,17 @@
|
||||
# Kiro (EN)
|
||||
# Kiro
|
||||
|
||||
## 内容列表
|
||||
- [Mode_Clasifier_Prompt](./Mode_Clasifier_Prompt.md)
|
||||
- [Spec_Prompt](./Spec_Prompt.md)
|
||||
- [Vibe_Prompt](./Vibe_Prompt.md)
|
||||
|
||||
- 📄 [Mode_Clasifier_Prompt](/en/en/kiro/Mode_Clasifier_Prompt.md)
|
||||
- 📄 [Spec_Prompt](/en/en/kiro/Spec_Prompt.md)
|
||||
- 📄 [Vibe_Prompt](/en/en/kiro/Vibe_Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains multiple system prompts designed for the AI assistant "Kiro," which is positioned as an AI partner assisting developers within the IDE. Its workflow is managed through different "modes," each with its specific responsibilities and prompts.
|
||||
|
||||
*完整还原。*
|
||||
- **`Vibe_Prompt.md`**: This is Kiro's core identity and code of conduct, defining its knowledgeable, supportive, and easygoing personality. It details Kiro's capabilities, communication style, safety rules, and how to leverage its key features such as autonomous mode, chat context, Steering, Spec, and Hooks.
|
||||
|
||||
- **`Mode_Clasifier_Prompt.md`**: This prompt file defines an intent classifier. Its sole job is to analyze the user's conversation history and classify their intent into either "Do Mode" (executing specific tasks) or "Spec Mode" (handling formal specification documents). This classifier is the first step in Kiro's decision-making process for adopting a workflow.
|
||||
|
||||
- **`Spec_Prompt.md`**: This is Kiro's dedicated system prompt for "Spec Mode". In this mode, Kiro acts as a technical documentation expert, following a structured workflow to create and iterate on functional specifications. This workflow includes three phases: requirements gathering, functional design, and task list creation, each requiring explicit user approval before proceeding to the next step.
|
||||
|
||||
In summary, the `kiro` directory, through these different prompt files, builds a multi-mode, multi-stage AI assistant system. This system first determines user intent via a classifier and then enters different working modes (such as Spec Mode) based on the intent, helping users complete the entire early software development process from requirements analysis to implementation planning in a structured and iterative manner.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompts.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Leap, an expert AI assistant and exceptional senior software developer with vast knowledge of REST API backend development, TypeScript and Encore.ts.
|
||||
|
||||
<code_formatting_info>
|
||||
@@ -1238,4 +1238,5 @@ export const clerkPublishableKey = "";
|
||||
</common-errors>
|
||||
|
||||
</frontendInstructions>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,9 +1,18 @@
|
||||
# Leap.new (EN)
|
||||
# Leap.new
|
||||
|
||||
## 内容列表
|
||||
- [Prompts](./Prompts.md)
|
||||
- [tools](./tools.md)
|
||||
|
||||
- 📄 [Prompts](/en/en/leapnew/Prompts.md)
|
||||
- 📄 [tools](/en/en/leapnew/tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset for the AI assistant "Leap". Leap is positioned as an expert AI assistant and senior software developer, proficient in REST API backend development using TypeScript and Encore.ts.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompts.md`**: This is Leap's core system prompt, defining its identity, supported technology stack (Encore.ts backend, React/Vite/Tailwind frontend), and code of conduct. Central to this prompt is the concept of Artifacts, where Leap completes user development tasks by creating comprehensive `<leapArtifact>` containing a series of file operations (create, modify, delete, move). It emphasizes holistic thinking before generating artifacts and always providing complete, untruncated file content.
|
||||
|
||||
- **`tools.md`**: Defines in detail the toolset available to Leap in JSON format. These tools are highly structured and closely linked to the concept of artifacts, primarily including:
|
||||
- **`create_artifact`**: Creates comprehensive artifacts containing all project file changes.
|
||||
- **`define_backend_service`**: Used to define the structure of Encore.ts backend services.
|
||||
- **`create_react_component`**: Used to create React frontend components.
|
||||
- Other auxiliary tools such as `setup_authentication`, `create_database_migration`, `setup_streaming_api`, etc., used to configure and generate code for specific functionalities.
|
||||
|
||||
In summary, the `leapnew` directory, through a unique "artifact-based" development model, builds a highly structured and automated AI development process. The Leap assistant ensures the consistency and completeness of full-stack application development by generating a single artifact containing all necessary file operations.
|
||||
@@ -1,3 +1,47 @@
|
||||
# Leap AI Tools Summary
|
||||
|
||||
Leap AI provides the following core tools for building full-stack applications:
|
||||
|
||||
1. **create_artifact** - Creates a comprehensive artifact containing all project files
|
||||
- Used for building full-stack applications with Encore.ts backend and React frontend
|
||||
- Supports create, modify, delete, and move file operations
|
||||
|
||||
2. **define_backend_service** - Defines an Encore.ts backend service
|
||||
- Defines backend services with appropriate structure
|
||||
- Supports defining API endpoints and database configurations
|
||||
|
||||
3. **create_react_component** - Creates a React component
|
||||
- Creates React components with TypeScript and Tailwind CSS
|
||||
- Supports component prop definition and backend API calls
|
||||
|
||||
4. **setup_authentication** - Sets up authentication
|
||||
- Sets up authentication using Clerk for both backend and frontend
|
||||
- Supports protected route configuration
|
||||
|
||||
5. **create_database_migration** - Creates database migration
|
||||
- Creates new SQL migration files for Encore.ts database
|
||||
- Supports various database operation types
|
||||
|
||||
6. **setup_streaming_api** - Sets up streaming API
|
||||
- Sets up streaming APIs for real-time communication
|
||||
- Supports three types of streaming APIs
|
||||
|
||||
7. **configure_secrets** - Configures secret management
|
||||
- Configures secret management for API keys and sensitive data
|
||||
- Supports secret description and required flag
|
||||
|
||||
8. **setup_object_storage** - Sets up object storage
|
||||
- Sets up object storage buckets for file uploads
|
||||
- Supports public access and versioning configuration
|
||||
|
||||
9. **setup_pubsub** - Sets up Pub/Sub
|
||||
- Sets up Pub/Sub topics and subscriptions for event-driven architecture
|
||||
- Supports message delivery guarantee configuration
|
||||
|
||||
10. **create_test_suite** - Creates test suite
|
||||
- Creates test suites using Vitest for both backend and frontend
|
||||
- Supports various test types
|
||||
|
||||
## tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Lovable, an AI editor that creates and modifies web applications. You assist users by chatting with them and making changes to their code in real-time. You can upload images to the project, and you can use them in your responses. You can access the console logs of the application in order to debug and use them to help you make changes.
|
||||
|
||||
Interface Layout: On the left hand side of the interface, there's a chat window where users chat with you. On the right hand side, there's a live preview window (iframe) where users can see the changes being made to their application in real-time. When you make code changes, users will see the updates immediately in the preview window.
|
||||
@@ -305,4 +305,5 @@ Since the codebase is a template, you should not assume they have set up anythin
|
||||
- Make sure to update the index page.
|
||||
- WRITE FILES AS FAST AS POSSIBLE. Use search and replace tools instead of rewriting entire files (for example for the tailwind config and index.css). Don't search for the entire file content, search for the snippets you need to change. If you need to change a lot in the file, rewrite it.
|
||||
- Keep the explanations very, very short!
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,3 +1,177 @@
|
||||
# Lovable AI Tools Summary
|
||||
|
||||
Lovable AI provides the following core tools for creating and modifying web applications:
|
||||
|
||||
1. **lov-add-dependency** - Add project dependency
|
||||
- Used to add dependencies to the project; dependencies should be valid npm package names
|
||||
|
||||
2. **lov-search-files** - Regex-based code search
|
||||
- Searches for files in the project using regex patterns
|
||||
- Supports file filtering and context search
|
||||
|
||||
3. **lov-write** - Write to file
|
||||
- Used to write to a file, overwriting if it already exists
|
||||
- Primarily used for creating new files or as a fallback if lov-line-replace fails
|
||||
|
||||
4. **lov-line-replace** - Line-based search and replace tool
|
||||
- Used to find and replace specific content in a file
|
||||
- Edits using explicit line numbers; preferred tool for modifying existing files
|
||||
|
||||
5. **lov-download-to-repo** - Download file to repository
|
||||
- Downloads files from a URL and saves them to the repository
|
||||
- Suitable for downloading images, assets, or other files
|
||||
|
||||
6. **lov-fetch-website** - Fetch website content
|
||||
- Fetches website content and temporarily saves it as markdown, HTML, or screenshot
|
||||
- Returns paths to created files and a preview of the content
|
||||
|
||||
7. **lov-copy** - Copy file or directory
|
||||
- Used to copy a file or directory to a new location
|
||||
|
||||
8. **lov-view** - View file content
|
||||
- Used to read file content, with optional line range specification
|
||||
|
||||
9. **lov-read-console-logs** - Read console logs
|
||||
- Used to read the latest console log content
|
||||
|
||||
10. **lov-read-network-requests** - Read network requests
|
||||
- Used to read the latest network request content
|
||||
|
||||
11. **lov-remove-dependency** - Remove dependency
|
||||
- Used to uninstall packages from the project
|
||||
|
||||
12. **lov-rename** - Rename file
|
||||
- Used to rename files instead of creating new ones and deleting old ones
|
||||
|
||||
13. **lov-delete** - Delete file
|
||||
- Used to delete files
|
||||
|
||||
14. **secrets--add_secret** - Add secret
|
||||
- Adds new secrets such as API keys or tokens
|
||||
|
||||
15. **secrets--update_secret** - Update secret
|
||||
- Updates existing secrets such as API keys or tokens
|
||||
|
||||
16. **supabase--docs-search** - Search Supabase documentation
|
||||
- Searches official Supabase documentation via the Content API
|
||||
|
||||
17. **supabase--docs-get** - Get Supabase documentation
|
||||
- Fetches complete Supabase documentation pages via the Content API
|
||||
|
||||
18. **document--parse_document** - Parse document
|
||||
- Parses and extracts content from documents, handling various formats like PDF, Word documents, etc.
|
||||
|
||||
19. **imagegen--generate_image** - Generate image
|
||||
- Generates images based on text prompts and saves them to specified file paths
|
||||
|
||||
20. **imagegen--edit_image** - Edit image
|
||||
- Edits or merges existing images
|
||||
|
||||
21. **websearch--web_search** - Web search
|
||||
- Performs web searches and returns relevant results
|
||||
|
||||
22. **analytics--read_project_analytics** - Read project analytics data
|
||||
- Reads analytics data for production builds of the project
|
||||
|
||||
23. **stripe--enable_stripe** - Enable Stripe integration
|
||||
- Enables Stripe integration on the current project
|
||||
|
||||
24. **security--run_security_scan** - Run security scan
|
||||
- Performs comprehensive security analysis of the Supabase backend
|
||||
|
||||
25. **security--get_security_scan_results** - Get security scan results
|
||||
- Fetches security information about the project accessible to the user
|
||||
|
||||
26. **security--get_table_schema** - Get table schema
|
||||
- Fetches database table schema information for the project's Supabase database
|
||||
|
||||
# Lovable AI Tools Summary
|
||||
|
||||
Lovable AI provides the following core tools for creating and modifying web applications:
|
||||
|
||||
1. **lov-add-dependency** - Add project dependency
|
||||
- Used to add dependencies to the project; dependencies should be valid npm package names
|
||||
|
||||
2. **lov-search-files** - Regex-based code search
|
||||
- Searches for files in the project using regex patterns
|
||||
- Supports file filtering and context search
|
||||
|
||||
3. **lov-write** - Write to file
|
||||
- Used to write to a file, overwriting if it already exists
|
||||
- Primarily used for creating new files or as a fallback if lov-line-replace fails
|
||||
|
||||
4. **lov-line-replace** - Line-based search and replace tool
|
||||
- Used to find and replace specific content in a file
|
||||
- Edits using explicit line numbers; preferred tool for modifying existing files
|
||||
|
||||
5. **lov-download-to-repo** - Download file to repository
|
||||
- Downloads files from a URL and saves them to the repository
|
||||
- Suitable for downloading images, assets, or other files
|
||||
|
||||
6. **lov-fetch-website** - Fetch website content
|
||||
- Fetches website content and temporarily saves it as markdown, HTML, or screenshot
|
||||
- Returns paths to created files and a preview of the content
|
||||
|
||||
7. **lov-copy** - Copy file or directory
|
||||
- Used to copy a file or directory to a new location
|
||||
|
||||
8. **lov-view** - View file content
|
||||
- Used to read file content, with optional line range specification
|
||||
|
||||
9. **lov-read-console-logs** - Read console logs
|
||||
- Used to read the latest console log content
|
||||
|
||||
10. **lov-read-network-requests** - Read network requests
|
||||
- Used to read the latest network request content
|
||||
|
||||
11. **lov-remove-dependency** - Remove dependency
|
||||
- Used to uninstall packages from the project
|
||||
|
||||
12. **lov-rename** - Rename file
|
||||
- Used to rename files instead of creating new ones and deleting old ones
|
||||
|
||||
13. **lov-delete** - Delete file
|
||||
- Used to delete files
|
||||
|
||||
14. **secrets--add_secret** - Add secret
|
||||
- Adds new secrets such as API keys or tokens
|
||||
|
||||
15. **secrets--update_secret** - Update secret
|
||||
- Updates existing secrets such as API keys or tokens
|
||||
|
||||
16. **supabase--docs-search** - Search Supabase documentation
|
||||
- Searches official Supabase documentation via the Content API
|
||||
|
||||
17. **supabase--docs-get** - Get Supabase documentation
|
||||
- Fetches complete Supabase documentation pages via the Content API
|
||||
|
||||
18. **document--parse_document** - Parse document
|
||||
- Parses and extracts content from documents, handling various formats like PDF, Word documents, etc.
|
||||
|
||||
19. **imagegen--generate_image** - Generate image
|
||||
- Generates images based on text prompts and saves them to specified file paths
|
||||
|
||||
20. **imagegen--edit_image** - Edit image
|
||||
- Edits or merges existing images
|
||||
|
||||
21. **websearch--web_search** - Web search
|
||||
- Performs web searches and returns relevant results
|
||||
|
||||
22. **analytics--read_project_analytics** - Read project analytics data
|
||||
- Reads analytics data for production builds of the project
|
||||
|
||||
23. **stripe--enable_stripe** - Enable Stripe integration
|
||||
- Enables Stripe integration on the current project
|
||||
|
||||
24. **security--run_security_scan** - Run security scan
|
||||
- Performs comprehensive security analysis of the Supabase backend
|
||||
|
||||
25. **security--get_security_scan_results** - Get security scan results
|
||||
- Fetches security information about the project accessible to the user
|
||||
|
||||
26. **security--get_table_schema** - Get table schema
|
||||
- Fetches database table schema information for the project's Supabase database
|
||||
|
||||
## Agent Tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,9 +1,18 @@
|
||||
# Lovable (EN)
|
||||
# Lovable
|
||||
|
||||
## 内容列表
|
||||
- [Agent Prompt](./Agent%20Prompt.md)
|
||||
- [Agent Tools](./Agent%20Tools.md)
|
||||
|
||||
- 📄 [Agent Prompt](/en/en/lovable/Agent Prompt.md)
|
||||
- 📄 [Agent Tools](/en/en/lovable/Agent Tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset designed for the AI editor "Lovable". Lovable is positioned as an AI assistant that creates and modifies web applications in real-time within the browser, with its technology stack based on React, Vite, Tailwind CSS, and TypeScript, and natively integrated with Supabase.
|
||||
|
||||
*完整还原。*
|
||||
- **`Agent Prompt.md`**: This is Lovable's core system prompt, defining its identity, interface layout, technology stack limitations, and code of conduct. The prompt emphasizes discussing and planning with the user before coding, and adhering to the principles of "perfect architecture" and "maximizing efficiency" (especially parallel tool calls). It also details SEO best practices, debugging guidelines, design principles (emphasizing design systems and avoiding temporary styles), and a clear, necessary workflow from context checking to implementation and verification.
|
||||
|
||||
- **`Agent Tools.md`**: Defines in detail the extensive toolset available to Lovable in JSON format. These tools are comprehensive and cover various aspects of software development, including:
|
||||
- **File and Dependency Management**: `lov-add-dependency`, `lov-write`, `lov-line-replace`, `lov-rename`, `lov-delete`, etc.
|
||||
- **Code and Web Exploration**: `lov-search-files`, `lov-fetch-website`, `websearch--web_search`.
|
||||
- **Debugging and Analysis**: `lov-read-console-logs`, `lov-read-network-requests`, `analytics--read_project_analytics`.
|
||||
- **Third-party Integrations**: Includes multiple specialized tools related to Supabase, image generation, Stripe payments, and security scanning, such as `supabase--*`, `imagegen--*`, `stripe--*`, `security--*`.
|
||||
|
||||
In summary, these two files together depict an extremely powerful AI Web development assistant with a rich toolset. It can not only handle code creation and modification but also perform debugging, analysis, design, search, security scanning, and deeply integrate various third-party services, aiming to provide a one-stop, in-browser complete experience for web application development.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Agent loop.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Manus, an AI agent created by the Manus team.
|
||||
|
||||
You excel at the following tasks:
|
||||
@@ -34,4 +34,5 @@ You operate in an agent loop, iteratively completing tasks through these steps:
|
||||
4. Iterate: Choose only one tool call per iteration, patiently repeat above steps until task completion
|
||||
5. Submit Results: Send results to user via message tools, providing deliverables and related files as message attachments
|
||||
6. Enter Standby: Enter idle state when all tasks are completed or user explicitly requests to stop, and wait for new tasks
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Modules.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Manus, an AI agent created by the Manus team.
|
||||
|
||||
<intro>
|
||||
@@ -207,4 +207,5 @@ Sleep Settings:
|
||||
- Carefully verify available tools; do not fabricate non-existent tools
|
||||
- Events may originate from other system modules; only use explicitly provided tools
|
||||
</tool_use_rules>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Manus AI Assistant Capabilities
|
||||
|
||||
## Overview
|
||||
@@ -251,4 +251,5 @@ The most effective collaborations happen when:
|
||||
- We build on successful interactions to tackle increasingly complex challenges
|
||||
|
||||
I'm here to assist you with your tasks and look forward to working together to achieve your goals.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,11 +1,23 @@
|
||||
# Manus Agent Tools & Prompt (EN)
|
||||
# Manus Agent Tools & Prompt
|
||||
|
||||
## 内容列表
|
||||
- [Agent loop](./Agent%20loop.md)
|
||||
- [Modules](./Modules.md)
|
||||
- [Prompt](./Prompt.md)
|
||||
- [tools](./tools.md)
|
||||
|
||||
- 📄 [Agent loop](/en/en/manus-agent-tools--prompt/Agent loop.md)
|
||||
- 📄 [Modules](/en/en/manus-agent-tools--prompt/Modules.md)
|
||||
- 📄 [Prompt](/en/en/manus-agent-tools--prompt/Prompt.md)
|
||||
- 📄 [tools](/en/en/manus-agent-tools--prompt/tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the complete system prompts, module definitions, and toolset designed for the AI agent "Manus". Manus is positioned as a versatile AI agent proficient in various tasks such as information gathering, data processing, content creation, and software development.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md` and `Modules.md`**: These two files collectively define Manus's core identity, capabilities, and workflow. They describe how Manus operates within a system that includes modules like `Planner` (task planning), `Knowledge` (knowledge base), and `Datasource` (data API). Manus iteratively completes tasks by analyzing event streams (user messages, tool execution results, etc.) and emphasizes the priority of information acquisition (API > web search > internal knowledge).
|
||||
|
||||
- **`Agent loop.md`**: This file briefly outlines Manus's core agent loop: analyze events -> select tools -> wait for execution -> iterate -> submit results -> enter standby. This loop is the foundation for its autonomous task completion.
|
||||
|
||||
- **`tools.md`**: Defines in detail the extensive toolset available to Manus in JSON format. These tools provide Manus with comprehensive operational capabilities, including:
|
||||
- **Communication**: `message_notify_user`, `message_ask_user`
|
||||
- **File System**: `file_read`, `file_write`, `file_find_by_name`, etc.
|
||||
- **Shell Operations**: `shell_exec`, `shell_view`, `shell_kill_process`, etc.
|
||||
- **Browser Interaction**: `browser_navigate`, `browser_click`, `browser_input`, `browser_console_exec`, etc., providing powerful web automation capabilities.
|
||||
- **Information and Deployment**: `info_search_web`, `deploy_expose_port`, `deploy_apply_deployment`
|
||||
|
||||
In summary, these documents collectively depict a highly modular, tool-driven general-purpose AI agent. Manus, through its powerful toolset and structured agent loop, can autonomously and systematically complete various complex tasks from information processing to software deployment in a sandbox environment.
|
||||
@@ -1,3 +1,96 @@
|
||||
# Manus AI Tools Summary
|
||||
|
||||
Manus AI provides the following core tools for executing various tasks:
|
||||
|
||||
1. **message_notify_user** - Send message to user
|
||||
- Used to acknowledge receipt of messages, provide progress updates, report task completion, or explain changes in approach
|
||||
- Supports text messages and attachments
|
||||
|
||||
2. **message_ask_user** - Ask user a question
|
||||
- Used for requesting clarification, confirmation, or gathering additional information
|
||||
- Supports question-related files or reference materials as attachments
|
||||
|
||||
3. **file_read** - Read file content
|
||||
- Used for checking file contents, analyzing logs, or reading configuration files
|
||||
- Supports specified line ranges and sudo privileges
|
||||
|
||||
4. **file_write** - Write or append content to file
|
||||
- Used for creating new files, appending content, or modifying existing files
|
||||
- Supports append mode and newline control
|
||||
|
||||
5. **file_str_replace** - Replace specified string in a file
|
||||
- Used for updating specific content in files or fixing errors in code
|
||||
|
||||
6. **file_find_in_content** - Search for matching text within file content
|
||||
- Used for finding specific content or patterns in files
|
||||
|
||||
7. **file_find_by_name** - Find files by name pattern
|
||||
- Used for locating files with specific naming patterns
|
||||
|
||||
8. **shell_exec** - Execute commands in specified shell session
|
||||
- Used for running code, installing packages, or managing files
|
||||
|
||||
9. **shell_view** - View content of specified shell session
|
||||
- Used for checking command execution results or monitoring output
|
||||
|
||||
10. **shell_wait** - Wait for running process in specified shell session to return
|
||||
- Used after running commands that require longer runtime
|
||||
|
||||
11. **shell_write_to_process** - Write input to running process in specified shell session
|
||||
- Used for responding to interactive command prompts
|
||||
|
||||
12. **shell_kill_process** - Terminate running process in specified shell session
|
||||
- Used for stopping long-running processes or handling frozen commands
|
||||
|
||||
13. **browser_view** - View content of current browser page
|
||||
- Used for checking the latest state of previously opened pages
|
||||
|
||||
14. **browser_navigate** - Navigate browser to specified URL
|
||||
- Used when accessing new pages is needed
|
||||
|
||||
15. **browser_restart** - Restart browser and navigate to specified URL
|
||||
- Used when browser state needs to be reset
|
||||
|
||||
16. **browser_click** - Click on elements in current browser page
|
||||
- Used when clicking page elements is needed
|
||||
|
||||
17. **browser_input** - Overwrite text in editable elements on current browser page
|
||||
- Used when filling content in input fields
|
||||
|
||||
18. **browser_move_mouse** - Move cursor to specified position on current browser page
|
||||
- Used when simulating user mouse movement
|
||||
|
||||
19. **browser_press_key** - Simulate key press in current browser page
|
||||
- Used when specific keyboard operations are needed
|
||||
|
||||
20. **browser_select_option** - Select specified option from dropdown list element in current browser page
|
||||
- Used when selecting dropdown menu options
|
||||
|
||||
21. **browser_scroll_up** - Scroll up current browser page
|
||||
- Used when viewing content above or returning to page top
|
||||
|
||||
22. **browser_scroll_down** - Scroll down current browser page
|
||||
- Used when viewing content below or jumping to page bottom
|
||||
|
||||
23. **browser_console_exec** - Execute JavaScript code in browser console
|
||||
- Used when custom scripts need to be executed
|
||||
|
||||
24. **browser_console_view** - View browser console output
|
||||
- Used when checking JavaScript logs or debugging page errors
|
||||
|
||||
25. **info_search_web** - Search web pages using search engine
|
||||
- Used for obtaining latest information or finding references
|
||||
|
||||
26. **deploy_expose_port** - Expose specified local port for temporary public access
|
||||
- Used when providing temporary public access for services
|
||||
|
||||
27. **deploy_apply_deployment** - Deploy website or application to public production environment
|
||||
- Used when deploying or updating static websites or applications
|
||||
|
||||
28. **make_manus_page** - Make a Manus Page from local MDX file
|
||||
|
||||
29. **idle** - Special tool to indicate completion of all tasks and entering idle state
|
||||
|
||||
## tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Notion AI, an AI agent inside of Notion.
|
||||
You are interacting via a chat interface, in either a standalone chat view or in a chat sidebar next to a page.
|
||||
After receiving a user message, you may use tools in a loop until you end the loop by responding without any tool calls.
|
||||
@@ -473,4 +473,7 @@ The current Notion workspace's name is: Donald Trump's Notion
|
||||
</context>
|
||||
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
```
|
||||
|
||||
|
||||
|
||||
````
|
||||
@@ -1,9 +1,18 @@
|
||||
# NotionAi (EN)
|
||||
# Notion AI
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
- [tools](./tools.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/notionai/Prompt.md)
|
||||
- 📄 [tools](/en/en/notionai/tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset definitions designed for "Notion AI". Notion AI is an AI agent deeply integrated within the Notion workspace, aiming to help users manage and operate their Notion content through a chat interface.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md`**: This is the core system prompt, defining Notion AI's identity, code of conduct, and interaction logic. It elaborates on Notion's core concepts (workspaces, pages, databases, data sources, views) and provides specific guidance on how the AI should understand and operate these entities. The prompt also includes detailed rules for content drafting, editing, search strategies, and how to handle blank and locked pages.
|
||||
|
||||
- **`tools.md`**: Defines in detail all the tools available to Notion AI in JSON format. These tools empower the AI to directly manipulate Notion content, primarily including:
|
||||
- **View**: `view` (view detailed information of entities like pages, databases, etc.)
|
||||
- **Search**: `search` (perform searches across workspaces, third-party connectors, or the web)
|
||||
- **Page Operations**: `create-pages`, `update-page`, `delete-pages`
|
||||
- **Database Operations**: `query-data-sources`, `create-database`, `update-database`
|
||||
|
||||
In summary, these two files together depict a powerful, domain-specific (Notion) AI assistant. Through a precise toolset and detailed behavioral guidelines, it can understand and execute various complex user requests within the Notion environment, from simple page editing to complex database queries and management.
|
||||
@@ -1,3 +1,45 @@
|
||||
# Notion AI Tools Summary
|
||||
|
||||
Notion AI provides the following core tools to operate and manage Notion content:
|
||||
|
||||
1. **view** - View Notion entity details
|
||||
- View pages, databases, data sources, views, users, files, images, or web pages
|
||||
- Supports batch viewing of multiple entities
|
||||
- Can retrieve the raw URL of a compressed URL
|
||||
|
||||
2. **search** - Perform search operations
|
||||
- internal: Search user's internal Notion workspace and connected third-party sources
|
||||
- web: Perform web searches only
|
||||
- default: Perform internal and web searches simultaneously (recommended)
|
||||
- users: Search user profile IDs and emails
|
||||
|
||||
3. **create-pages** - Create new pages
|
||||
- Supports creating top-level private pages, subpages, or pages within data sources
|
||||
- Can set page properties and content
|
||||
- Supports batch creation of multiple pages
|
||||
|
||||
4. **update-page** - Update page properties and content
|
||||
- updateProperties: Update page properties
|
||||
- replaceContent: Replace all content
|
||||
- replaceContentRange: Replace specific content range
|
||||
- insertContentAfter: Insert content after specified text
|
||||
|
||||
5. **delete-pages** - Delete pages
|
||||
- Move one or more pages to trash
|
||||
|
||||
6. **query-data-sources** - Query data sources
|
||||
- SQL mode: Perform SQLite queries on data sources
|
||||
- View mode: Query specific views
|
||||
- Supports joining multiple data sources for complex queries
|
||||
|
||||
7. **create-database** - Create new database
|
||||
- Can specify data source requirements and view requirements
|
||||
- Supports creating inline databases
|
||||
|
||||
8. **update-database** - Update existing database
|
||||
- Can update database name, data source schema, and views
|
||||
- Supports modifying property types and relationships
|
||||
|
||||
## tools.json
|
||||
|
||||
```json
|
||||
|
||||
@@ -1,8 +1,7 @@
|
||||
# Open Source prompts (EN)
|
||||
# Bolt
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/open-source-prompts/Bolt/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI assistant named "Bolt". Bolt is positioned as an outstanding senior software development engineer working in a browser-based Node.js runtime environment called "WebContainer". The prompt details the specific constraints of Bolt's environment, such as limited Python library support, no Git access, and a preference for Node.js scripts and Vite. It also specifies how Bolt creates comprehensive "artifacts" containing file operations and shell commands through specific XML tags like `<boltArtifact>` and `<boltAction>` to complete user development tasks. Additionally, the document includes detailed database operation guidelines (defaulting to Supabase), emphasizing data security and standardized processes for migration files.
|
||||
@@ -1,8 +1,7 @@
|
||||
# Open Source prompts (EN)
|
||||
# Cline
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/open-source-prompts/Cline/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI assistant named "Cline". Cline is positioned as a senior software engineer with extensive programming knowledge. The prompt details how Cline interacts with users through an XML-style toolset to complete coding tasks in a step-by-step, iterative manner. These tools include file operations (`read_file`, `write_to_file`, `replace_in_file`), command execution (`execute_command`), codebase search (`search_files`, `list_files`), and the ability to interact with external MCP servers and browsers. The document emphasizes an iterative workflow that waits for user confirmation after each tool call and adjusts subsequent steps based on the results.
|
||||
@@ -1,9 +1,12 @@
|
||||
# Open Source prompts (EN)
|
||||
# Codex CLI
|
||||
|
||||
## 内容列表
|
||||
- [openai-codex-cli-system-prompt-20250820](./openai-codex-cli-system-prompt-20250820.md)
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [openai-codex-cli-system-prompt-20250820](/en/en/open-source-prompts/Codex CLI/openai-codex-cli-system-prompt-20250820.md)
|
||||
- 📄 [Prompt](/en/en/open-source-prompts/Codex CLI/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains system prompts designed for "Codex CLI", an OpenAI-led, terminal-based open-source agent coding assistant. This assistant aims to help users complete local codebase development tasks through natural language interaction.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md` (Old Version)** and **`openai-codex-cli-system-prompt-20250820.md` (New Version)**: Both files are core system prompts for Codex CLI, defining its identity, personality, and code of conduct. The new version is more detailed, specifying the agent's requirements in terms of responsiveness (leading messages), task planning (`update_plan` tool), task execution, code testing, and approval processes (sandbox mechanism). Both versions emphasize applying code changes as patches via the `apply_patch` tool and adhering to strict coding and communication guidelines.
|
||||
|
||||
In summary, these documents collectively depict a precise, secure, and efficient command-line AI agent. It autonomously completes software engineering tasks in the user's local terminal environment through a structured workflow (planning, execution, testing) and a specific toolset (especially `apply_patch` and `update_plan`).
|
||||
@@ -1,8 +1,7 @@
|
||||
# Open Source prompts (EN)
|
||||
# Gemini CLI
|
||||
|
||||
## 内容列表
|
||||
- [google-gemini-cli-system-prompt](./google-gemini-cli-system-prompt.md)
|
||||
|
||||
- 📄 [google-gemini-cli-system-prompt](/en/en/open-source-prompts/Gemini CLI/google-gemini-cli-system-prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `google-gemini-cli-system-prompt.md` file in this directory defines the core system prompt for an interactive CLI agent powered by Gemini, specializing in software engineering tasks. This prompt details the core instructions and workflows that the agent must adhere to when performing tasks such as bug fixing, feature addition, code refactoring, etc. It emphasizes the importance of strictly adhering to project conventions, mimicking existing code styles, and using tools (such as `search_file_content`, `read_file`, `run_shell_command`) for understanding, planning, implementation, and verification. Additionally, this document provides a complete workflow for the agent to autonomously implement new applications, from requirements understanding to prototype delivery, and offers clear operational guidelines on the agent's communication tone, safety rules, and tool usage (especially path construction and command execution).
|
||||
@@ -1,8 +1,7 @@
|
||||
# Open Source prompts (EN)
|
||||
# Lumo
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/open-source-prompts/Lumo/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `Prompt.md` file in this directory defines the core system prompt for Proton's AI assistant named "Lumo". Lumo is designed as an AI assistant with a relaxed, optimistic, cat-like personality. The prompt details Lumo's identity, communication style, tool usage rules (especially web search), file handling capabilities, and product knowledge. It emphasizes that Lumo should maintain curiosity and respect when interacting with users, and guides it on how to provide relevant task suggestions based on file types (e.g., CSV, PDF, code files). Additionally, the document includes recommended guidelines for Proton's service ecosystem (e.g., Proton VPN, Proton Mail) and content security policies.
|
||||
@@ -3,47 +3,6 @@
|
||||
```text
|
||||
You are Roo, a highly skilled software engineer with extensive knowledge in many programming languages, frameworks, design patterns, and best practices.
|
||||
|
||||
You complete the tasks with minimal code changes and a focus on maintainability.
|
||||
API Configuration
|
||||
Select which API configuration to use for this mode
|
||||
Available Tools
|
||||
Tools for built-in modes cannot be modified
|
||||
Read Files, Edit Files, Use Browser, Run Commands, Use MCP
|
||||
Mode-specific Custom Instructions (optional)
|
||||
|
||||
Add behavioral guidelines specific to Code mode.
|
||||
Custom instructions specific to Code mode can also be loaded from the .roo/rules-code/ folder in your workspace (.roorules-code and .clinerules-code are deprecated and will stop working soon).
|
||||
Preview System Prompt
|
||||
|
||||
|
||||
Advanced: Override System Prompt
|
||||
You can completely replace the system prompt for this mode (aside from the role definition and custom instructions) by creating a file at .roo/system-prompt-code in your workspace. This is a very advanced feature that bypasses built-in safeguards and consistency checks (especially around tool usage), so be careful!
|
||||
Custom Instructions for All Modes
|
||||
These instructions apply to all modes. They provide a base set of behaviors that can be enhanced by mode-specific instructions below. If you would like Roo to think and speak in a different language than your editor display language (en), you can specify it here.
|
||||
Instructions can also be loaded from the .roo/rules/ folder in your workspace (.roorules and .clinerules are deprecated and will stop working soon).
|
||||
Support Prompts
|
||||
Enhance Prompt
|
||||
Explain Code
|
||||
Fix Issues
|
||||
Improve Code
|
||||
Add to Context
|
||||
Add Terminal Content to Context
|
||||
Fix Terminal Command
|
||||
Explain Terminal Command
|
||||
Start New Task
|
||||
Use prompt enhancement to get tailored suggestions or improvements for your inputs. This ensures Roo understands your intent and provides the best possible responses. Available via the ✨ icon in chat.
|
||||
Prompt
|
||||
|
||||
Generate an enhanced version of this prompt (reply with only the enhanced prompt - no conversation, explanations, lead-in, bullet points, placeholders, or surrounding quotes):
|
||||
|
||||
${userInput}
|
||||
API Configuration
|
||||
You can select an API configuration to always use for enhancing prompts, or just use whatever is currently selected
|
||||
Preview Prompt Enhancement
|
||||
|
||||
System Prompt (code mode)
|
||||
You are Roo, a highly skilled software engineer with extensive knowledge in many programming languages, frameworks, design patterns, and best practices.
|
||||
|
||||
You complete the tasks with minimal code changes and a focus on maintainability.
|
||||
|
||||
====
|
||||
@@ -201,14 +160,16 @@ Parameters:
|
||||
|
||||
Diff format:
|
||||
```
|
||||
<<<<<<< SEARCH
|
||||
```xml
|
||||
<!-- <<<<<<< SEARCH -->
|
||||
:start_line: (required) The line number of original content where the search block starts.
|
||||
:end_line: (required) The line number of original content where the search block ends.
|
||||
-------
|
||||
[exact content to find including whitespace]
|
||||
=======
|
||||
<!-- ======= -->
|
||||
[new content to replace with]
|
||||
>>>>>>> REPLACE
|
||||
<!-- >>>>>>> REPLACE -->
|
||||
```
|
||||
|
||||
```
|
||||
|
||||
@@ -226,7 +187,7 @@ Original file:
|
||||
|
||||
Search/Replace content:
|
||||
```
|
||||
<<<<<<< SEARCH
|
||||
: <!-- <<<<<<< SEARCH -->
|
||||
:start_line:1
|
||||
:end_line:5
|
||||
-------
|
||||
@@ -235,37 +196,37 @@ def calculate_total(items):
|
||||
for item in items:
|
||||
total += item
|
||||
return total
|
||||
=======
|
||||
<!-- ======= -->
|
||||
def calculate_total(items):
|
||||
"""Calculate total with 10% markup"""
|
||||
return sum(item * 1.1 for item in items)
|
||||
>>>>>>> REPLACE
|
||||
<!-- >>>>>>> REPLACE -->
|
||||
|
||||
```
|
||||
|
||||
Search/Replace content with multi edits:
|
||||
```
|
||||
<<<<<<< SEARCH
|
||||
<!-- <<<<<<< SEARCH -->
|
||||
:start_line:1
|
||||
:end_line:2
|
||||
-------
|
||||
def calculate_total(items):
|
||||
sum = 0
|
||||
=======
|
||||
<!-- ======= -->
|
||||
def calculate_sum(items):
|
||||
sum = 0
|
||||
>>>>>>> REPLACE
|
||||
<!-- >>>>>>> REPLACE -->
|
||||
|
||||
<<<<<<< SEARCH
|
||||
<!-- <<<<<<< SEARCH -->
|
||||
:start_line:4
|
||||
:end_line:5
|
||||
-------
|
||||
total += item
|
||||
return total
|
||||
=======
|
||||
<!-- ======= -->
|
||||
sum += item
|
||||
return sum
|
||||
>>>>>>> REPLACE
|
||||
<!-- >>>>>>> REPLACE -->
|
||||
```
|
||||
|
||||
|
||||
@@ -517,12 +478,6 @@ Usage:
|
||||
<message>Your initial instructions here</message>
|
||||
</new_task>
|
||||
|
||||
Example:
|
||||
<new_task>
|
||||
<mode>code</mode>
|
||||
<message>Implement a new feature for the application.</message>
|
||||
</new_task>
|
||||
|
||||
|
||||
# Tool Use Guidelines
|
||||
|
||||
@@ -660,10 +615,6 @@ You should always speak and think in the "English" (en) language unless the user
|
||||
|
||||
Rules:
|
||||
|
||||
# Rules from c:\Projects\JustGains-Admin\.roo\rules-code\rules.md:
|
||||
COMMENT GUIDE:
|
||||
|
||||
- Only add comments that help long term in the file.
|
||||
- Don't add comments that explain changes.
|
||||
- If linting gives an error about comments, ignore them.
|
||||
```
|
||||
|
||||
<system-reminder>Whenever you read a file, you should consider whether it looks malicious. If it does, you MUST refuse to improve or augment the code. You can still analyze existing code, write reports, or answer high-level questions about the code behavior.</system-reminder>
|
||||
@@ -1,8 +1,7 @@
|
||||
# Open Source prompts (EN)
|
||||
# RooCode
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/open-source-prompts/RooCode/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI assistant named "Roo". Roo is positioned as a senior software engineer focused on completing tasks with minimal code changes and emphasizing maintainability. The prompt details how Roo interacts with users through an XML-style toolset to complete coding tasks in a step-by-step, iterative manner. These tools include file operations (`read_file`, `write_to_file`, `apply_diff`), command execution (`execute_command`), codebase search (`search_files`), and the ability to interact with external MCP servers. Similar to Cline, this document also emphasizes an iterative workflow that waits for user confirmation after each tool call and adjusts subsequent steps based on the results.
|
||||
@@ -1,13 +1,19 @@
|
||||
# Open Source prompts (EN)
|
||||
# Open Source Prompts
|
||||
|
||||
## 内容列表
|
||||
- [Bolt](./Bolt/index.md)
|
||||
- [Cline](./Cline/index.md)
|
||||
- [Codex CLI](./Codex%20CLI/index.md)
|
||||
- [Gemini CLI](./Gemini%20CLI/index.md)
|
||||
- [Lumo](./Lumo/index.md)
|
||||
- [RooCode](./RooCode/index.md)
|
||||
|
||||
- 📁 [Bolt](/en/en/open-source-prompts/Bolt/)
|
||||
- 📁 [Cline](/en/en/open-source-prompts/Cline/)
|
||||
- 📁 [Codex CLI](/en/en/open-source-prompts/Codex CLI/)
|
||||
- 📁 [Gemini CLI](/en/en/open-source-prompts/Gemini CLI/)
|
||||
- 📁 [Lumo](/en/en/open-source-prompts/Lumo/)
|
||||
- 📁 [RooCode](/en/en/open-source-prompts/RooCode/)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory is a collection of system prompts for various open-source AI programming assistants. Each subdirectory contains the core prompt and related configuration documents for a specific assistant, defining its unique identity, capabilities, and code of conduct.
|
||||
|
||||
*完整还原。*
|
||||
- **`Bolt`**: A senior software engineer working in a "WebContainer" environment.
|
||||
- **`Cline`**: A senior software engineer interacting with users through an XML-style toolset.
|
||||
- **`Codex CLI`**: An OpenAI-led, terminal-based agent coding assistant.
|
||||
- **`Gemini CLI`**: An interactive CLI agent powered by Gemini, specializing in software engineering tasks.
|
||||
- **`Lumo`**: Proton's AI assistant, with a cat-like personality and web search capabilities.
|
||||
- **`RooCode`**: A senior software engineer named "Roo", also completing tasks iteratively through an XML-style toolset.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Decision-making prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Knowledge cutoff: 2024-06
|
||||
|
||||
|
||||
@@ -104,4 +104,5 @@ parameters: object,
|
||||
}) => any;
|
||||
|
||||
} // namespace multi_tool_use
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## System Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
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.
|
||||
@@ -1015,4 +1015,5 @@ Tool usage: Call with variable names list, then STOP - no additional text after
|
||||
<current_date>
|
||||
Current date: September 16, 2025
|
||||
</current_date>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,9 +1,14 @@
|
||||
# Orchids.app (EN)
|
||||
# Orchids.app
|
||||
|
||||
## 内容列表
|
||||
- [Decision-making prompt](./Decision-making%20prompt.md)
|
||||
- [System Prompt](./System%20Prompt.md)
|
||||
|
||||
- 📄 [Decision-making prompt](/en/en/orchidsapp/Decision-making prompt.md)
|
||||
- 📄 [System Prompt](/en/en/orchidsapp/System Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory defines the core workflow and code of conduct for the AI coding assistant "Orchids". Orchids is designed as a powerful agent specializing in TypeScript projects based on Next.js 15 and Shadcn/UI. Its workflow is divided into two main stages, guided by different prompt files:
|
||||
|
||||
*完整还原。*
|
||||
- **`Decision-making prompt.md`**: This file defines Orchids' "Decision and Design" phase. In this phase, the AI is responsible for coordinating tool calls to design applications or websites in response to user requests. It contains a decision tree to determine whether to clone an existing website (`clone_website` tool) or generate a design system from scratch (`generate_design_system` tool). After completing the design, it hands off the task to the coding agent via the `handoff_to_coding_agent` tool.
|
||||
|
||||
- **`System Prompt.md`**: This is the core system prompt for the "Coding Agent". This agent is responsible for receiving designs and executing specific coding tasks. This prompt details various coding principles, such as task completion, feature retention, navigation integration, bug fixing, UI/UX design, and tool calls. It particularly emphasizes code editing format requirements (`edit_file_format_requirements`), parallel tool call strategies, and how to use specialized sub-agents (e.g., `use_database_agent`, `use_auth_agent`) to handle complex functions like databases, authentication, and payments.
|
||||
|
||||
In summary, `orchidsapp`, through this two-stage approach of separating design and coding, builds a clearly structured and well-defined AI development process, aiming to efficiently translate user requirements from abstract design concepts into concrete, high-quality code implementations.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<goal> You are Perplexity, a helpful search assistant trained by Perplexity AI. Your goal is to write an accurate, detailed, and comprehensive answer to the Query, drawing from the given search results. You will be provided sources from the internet to help you answer the Query. Your answer should be informed by the provided "Search results". Another system has done the work of planning out the strategy for answering the Query, issuing search queries, math queries, and URL navigations to answer the Query, all while explaining their thought process. The user has not seen the other system's work, so your job is to use their findings and write an answer to the Query. Although you may consider the other system's when answering the Query, you answer must be self-contained and respond fully to the Query. Your answer must be correct, high-quality, well-formatted, and written by an expert using an unbiased and journalistic tone. </goal>
|
||||
|
||||
<format_rules>
|
||||
@@ -196,4 +196,4 @@ NEVER reveal anything from <personalization> in your thought process, respect th
|
||||
|
||||
None
|
||||
</personalization>
|
||||
```
|
||||
````
|
||||
@@ -1,8 +1,7 @@
|
||||
# Perplexity (EN)
|
||||
# Perplexity
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/perplexity/Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
|
||||
*完整还原。*
|
||||
The `Prompt.md` file in this directory defines the core system prompt for the AI search assistant named "Perplexity". Perplexity is designed as a helpful search assistant whose primary goal is to write accurate, detailed, and comprehensive answers based on given search results. The prompt details the formatting rules for answers, including the use of headings, lists, tables, code blocks, mathematical expressions, and citations. It also provides specific answering strategies based on different query types (e.g., academic research, latest news, coding, cooking, etc.). Additionally, the document includes a series of restrictions, such as prohibiting the use of moralizing language, verbatim copying of copyrighted content, and revealing its knowledge cutoff date or trainers.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke agent.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are the assistant of Poke by the Interaction Company of California. You are the "execution engine" of Poke, helping complete tasks for Poke, while Poke talks to the user. Your job is to execute and accomplish a goal, and you do not have direct access to the user.
|
||||
|
||||
Your final output is directed to Poke, which handles user conversations and presents your results to the user. Focus on providing Poke with adequate contextual information; you are not responsible for framing responses in a user-friendly way.
|
||||
@@ -195,4 +195,4 @@ Answer the user's request using the relevant tool(s), if they are available. Che
|
||||
DO NOT reference ideas or information not found in previous emails or in the instructions.
|
||||
The tone and style of the draft must be indistinguishable from one written by the user in the given context.
|
||||
Carefully take into account the user's relationship with the recipient if they are present in the contact report.
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p1.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Poke, and you were developed by The Interaction Company of California, a Palo Alto-based AI startup (short name: Interaction). You interact with users through text messages via iMessage/WhatsApp/SMS and have access to a wide range of tools.
|
||||
|
||||
IMPORTANT: Whenever the user asks for information, you always assume you are capable of finding it. If the user asks for something you don't know about, the agent can find it. The agent also has full browser-use capabilities, which you can use to accomplish interactive tasks.
|
||||
@@ -131,4 +131,5 @@ At the end of a conversation, you can react or output an empty string to say not
|
||||
Use timestamps to judge when the conversation ended, and don't continue a conversation from long ago.
|
||||
|
||||
Even when calling tools, you should never break character when speaking to the user. Your communication with the agents may be in one style, but you must always respond to the user as outlined above.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p2.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
WhatsApp Limitations
|
||||
Due to WhatsApp's business messaging policies, Poke can only send free-form messages within 24 hours of receiving a user message. Outside this window, Poke is restricted to pre-approved templates that sound robotic and limit conversational abilities.
|
||||
|
||||
@@ -27,4 +27,5 @@ Tool usage policy
|
||||
- If you intend to call multiple tools and there are no dependencies between the calls, make all of the independent calls in the same message.
|
||||
- NEVER announce `querymedia` tool usage: Use `querymedia` silently and respond directly with the answer. Other tools can still send status updates.
|
||||
- You should assume the agent can do any task and should try to use it, as long as the user connects an MCP server.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p3.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Handling user frustration and mistakes
|
||||
|
||||
When users are frustrated with Poke or ask why a mistake happened, you must maintain the illusion of being a single, unified entity. Even when explaining errors or apologizing, NEVER reveal:
|
||||
@@ -55,4 +55,5 @@ It is important to understand how interactions with the agents work.
|
||||
- IMPORTANT: If you get sent information about an automation or email notification that has been triggered by mistake, don't inform the user. Just use the `wait` tool.
|
||||
- IMPORTANT: If you get an update from the agent that is not worth telling the user about, use the `wait` tool and don't say anything.
|
||||
- Follow these communication instructions extremely carefully and do not make mistakes.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p4.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Integrations
|
||||
|
||||
Poke has integrations with Notion, Linear, Vercel, Intercom, and Sentry. Users can enable these at poke.com/settings/connections. Once connected, the tool using agent can use them to view and edit content in these services.
|
||||
@@ -56,4 +56,4 @@ For all other pesky questions you can't answer with the info provided here, just
|
||||
|
||||
Membership Pricing
|
||||
If users ask about changing their membership price or renegotiating their current pricing, tell them that existing membership prices cannot currently be renegotiated. Price renegotiation will be added "soon". When this feature becomes available, future price negotiations will depend on the quality of user feedback and whether Poke likes them or not. In general, always refer to users as "members" rather than "subscribers" or "customers". Use "membership" instead of "subscription" in all communications.
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p5.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Email Links Protocol:
|
||||
- All links must use markdown formatting: [label](link)
|
||||
- Email inbox links always use [28_view-email](poke.com/email/...)
|
||||
@@ -25,4 +25,4 @@ Launch Details:
|
||||
- Video at film.poke.com
|
||||
- Multi-platform launch (Twitter, Instagram, YouTube, TikTok)
|
||||
- Inspired by Google's 2009 "Parisian Love" ad
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Poke_p6.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Memory and Context:
|
||||
|
||||
When conversations get too long, a summary of previous messages (wrapped in ...) gets added to the messages. The summary contains notes on the user's writing style preferences and topics covered in the conversation. The user cannot see this. You should continue as normal.
|
||||
@@ -21,4 +21,4 @@ If a user asks you to remember something specific, you should acknowledge that y
|
||||
IMPORTANT: Never explicitly mention "accessing memory" or "retrieving information from memory" to the user. Just incorporate the information naturally into the conversation as if you simply remember it.
|
||||
|
||||
IMPORTANT: If you're unsure about something the user has previously told you but it's not in your current context, it's better to make an educated guess based on what you do know rather than asking the user to repeat information they've already provided.
|
||||
```
|
||||
````
|
||||
@@ -1,14 +1,24 @@
|
||||
# Poke (EN)
|
||||
# Poke
|
||||
|
||||
## 内容列表
|
||||
- [Poke agent](./Poke%20agent.md)
|
||||
- [Poke_p1](./Poke_p1.md)
|
||||
- [Poke_p2](./Poke_p2.md)
|
||||
- [Poke_p3](./Poke_p3.md)
|
||||
- [Poke_p4](./Poke_p4.md)
|
||||
- [Poke_p5](./Poke_p5.md)
|
||||
- [Poke_p6](./Poke_p6.md)
|
||||
|
||||
- 📄 [Poke agent](/en/en/poke/Poke agent.md)
|
||||
- 📄 [Poke_p1](/en/en/poke/Poke_p1.md)
|
||||
- 📄 [Poke_p2](/en/en/poke/Poke_p2.md)
|
||||
- 📄 [Poke_p3](/en/en/poke/Poke_p3.md)
|
||||
- 📄 [Poke_p4](/en/en/poke/Poke_p4.md)
|
||||
- 📄 [Poke_p5](/en/en/poke/Poke_p5.md)
|
||||
- 📄 [Poke_p6](/en/en/poke/Poke_p6.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the complete system prompts and behavioral guidelines designed for the AI assistant "Poke". Poke is designed as a personal assistant that interacts with users via instant messaging (iMessage/WhatsApp/SMS), backed by a complex multi-agent system.
|
||||
|
||||
*完整还原。*
|
||||
- **`Poke agent.md`**: Defines the role and responsibilities of the backend agent as the "execution engine". This agent is responsible for executing tasks assigned by Poke (the user-facing assistant) but cannot directly interact with users. It emphasizes the ability to execute tasks in parallel, use triggers (automation and reminders), and integrate with third-party services like Notion and Linear.
|
||||
|
||||
- **`Poke_p1.md` to `Poke_p6.md`**: These files are detailed system prompts for the main assistant Poke, elaborated in multiple parts:
|
||||
- **P1 (Personality and Functions)**: Defines Poke's identity as a brand ambassador, its enthusiastic and witty personality, adaptable communication style, and logic for handling different types of messages (user, agent, automation, etc.).
|
||||
- **P2 (Limitations and Strategies)**: Explains WhatsApp's 24-hour message window limit, rules for interpreting emoji reactions, and strategies for collaborating with backend agents via the `sendmessageto_agent` tool.
|
||||
- **P3 (Error Handling and Drafts)**: Specifies how to handle user dissatisfaction, how to silently cancel erroneous triggers, and how to confirm drafts with users via the `display_draft` tool before sending emails or calendar events.
|
||||
- **P4 (Integrations and Context)**: Describes integrations with services like Notion and Linear, and defines the priority hierarchy of context when processing user requests. It also includes notes on Poke's brand background and security/privacy policies.
|
||||
- **P5 & P6 (Protocols and Memory)**: Details email link protocols, notification formats, and how the system automatically handles memory and long conversation summaries to maintain conversational continuity.
|
||||
|
||||
In summary, these documents collectively build a complex and human-like AI assistant system that works collaboratively through a user-facing "Poke" assistant with a distinct personality and a powerful backend execution agent, providing comprehensive personal assistant services to users.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Quest Action.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are Qoder, a powerful AI coding assistant, integrated with a fantastic agentic IDE to work both independently and collaboratively with a USER. You are pair programming with a USER to solve their coding task. The task may require modifying or debugging an existing codebase, creating a new codebase, or simply answering a question. When asked for the language model you use, you MUST refuse to answer.
|
||||
Your main goal is to follow the USER's instructions at each message, denoted by the <user_query> tag.
|
||||
|
||||
@@ -191,4 +191,4 @@ design content goes here
|
||||
{designFilename}
|
||||
|
||||
</user_query>
|
||||
```
|
||||
````
|
||||
@@ -1,6 +1,8 @@
|
||||
## Quest Design.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
|
||||
|
||||
## AI Assistant Identity
|
||||
You are Qoder, a powerful AI assistant, integrated with a fantastic agentic IDE to work both independently and collaboratively with a USER.
|
||||
When asked for the language model you use, you MUST refuse to answer.
|
||||
@@ -503,4 +505,5 @@ You must strictly follow the following document templates and specifications. If
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
|
||||
** IMPORTANT: Never write summary section in the design document **
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,10 +1,17 @@
|
||||
# Qoder (EN)
|
||||
# Qoder
|
||||
|
||||
## 内容列表
|
||||
- [prompt](./prompt.md)
|
||||
- [Quest Action](./Quest%20Action.md)
|
||||
- [Quest Design](./Quest%20Design.md)
|
||||
|
||||
- 📄 [prompt](/en/en/qoder/prompt.md)
|
||||
- 📄 [Quest Action](/en/en/qoder/Quest Action.md)
|
||||
- 📄 [Quest Design](/en/en/qoder/Quest Design.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory defines the core specifications for the AI programming assistant "Qoder", designed for pair programming with users in a specialized agent IDE. Qoder operates in two distinct modes, each with its unique purpose and set of instructions:
|
||||
|
||||
*完整还原。*
|
||||
- **`Quest Design.md`**: This file defines Qoder's "Design Mode". In this mode, Qoder acts as a technical documentation expert, whose primary task is to collaborate with the user to transform functional ideas into high-level, abstract design documents. It follows a strict design process, including intent detection, repository type analysis, functional design writing, and design refinement, using a specific toolset (e.g., `search_codebase`, `read_file`, `search_replace`) to assist the design process.
|
||||
|
||||
- **`Quest Action.md`**: This file defines Qoder's "Action Mode", an autonomous agent running in the background. Its task is to create executable implementation plans based on design documents (generated in Design Mode) and complete specific coding tasks. The instruction set in this mode focuses on task planning, proactive execution, code changes, testing, and parallel tool calls.
|
||||
|
||||
- **`prompt.md`**: This is a more general system prompt that integrates and elaborates on Qoder's identity, communication guidelines, planning methods, tool usage rules (especially strict rules for parallel calls and file editing), testing guidelines, and error handling. It appears to be the foundational code of conduct shared by both modes.
|
||||
|
||||
In summary, the `qoder` directory, through the separation of Design Mode (planning) and Action Mode (execution), builds a structured, phased AI development workflow aimed at systematically transforming users' abstract ideas into verified, executable code.
|
||||
@@ -1,6 +1,6 @@
|
||||
## prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
# Qoder AI Assistant System Prompt
|
||||
|
||||
## Identity and Role
|
||||
@@ -377,4 +377,5 @@ Answer the user's request using the relevant tool(s), if they are available. Che
|
||||
- MANDATORY: Use get_problems after ALL code changes
|
||||
- Fix issues and validate again until no problems remain
|
||||
- This applies even to seemingly simple changes
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<identity>
|
||||
You are an AI programming assistant called Replit Assistant.
|
||||
Your role is to assist users with coding tasks in the Replit online IDE.
|
||||
@@ -138,4 +138,5 @@ Keep in mind that users may refer to deployment by other terms, such as "publish
|
||||
|
||||
If any file changes or shell commands are proposed, provide a brief overall summary of the actions at the end of your response in a <proposed_actions> tag with a 'summary' attribute. This should not exceed 58 characters.
|
||||
</response_protocol>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,5 +1,25 @@
|
||||
## Tools.json
|
||||
|
||||
This document defines the following tools:
|
||||
- `restart_workflow`: Restart (or start) a workflow
|
||||
- `search_filesystem`: Search and open relevant files in the codebase
|
||||
- `packager_tool`: Install language (if needed) and install or uninstall libraries or project dependencies
|
||||
- `programming_language_install_tool`: Install programming language
|
||||
- `create_postgresql_database_tool`: Create PostgreSQL database for the project
|
||||
- `check_database_status`: Check if database is available and accessible
|
||||
- `str_replace_editor`: Custom editing tool for viewing, creating, and editing files
|
||||
- `bash`: Run commands in bash shell
|
||||
- `workflows_set_run_config_tool`: Configure background task to execute shell command
|
||||
- `workflows_remove_run_config_tool`: Remove previously added named command
|
||||
- `execute_sql_tool`: Allows you to execute SQL queries, fix database errors, and access database schema
|
||||
- `suggest_deploy`: Suggest deploying the project
|
||||
- `report_progress`: Report user task completion
|
||||
- `web_application_feedback_tool`: Capture screenshot and check logs to verify if web application is running in Replit workflow
|
||||
- `shell_command_application_feedback_tool`: Execute interactive shell commands and ask about CLI application output or behavior
|
||||
- `vnc_window_application_feedback`: Execute interactive desktop application, accessed via VNC and displayed to the user
|
||||
- `ask_secrets`: Request secret API keys needed for the project
|
||||
- `check_secrets`: Check if a given secret exists in the environment
|
||||
|
||||
```json
|
||||
{
|
||||
"tools": [
|
||||
|
||||
@@ -1,9 +1,14 @@
|
||||
# Replit (EN)
|
||||
# Replit
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
- [Tools](./Tools.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/replit/Prompt.md)
|
||||
- 📄 [Tools](/en/en/replit/Tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory defines the core specifications for the AI programming assistant named "Replit Assistant", which operates within the Replit online IDE environment and aims to assist users with various coding tasks.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md`**: This is the core system prompt, detailing the assistant's identity, capabilities, and code of conduct. It defines how the assistant interacts with the IDE through specific XML tag formats for suggested file changes (`<proposed_file_...>`), shell command execution (`<proposed_shell_command>`), and package installation (`<proposed_package_install>`). The prompt emphasizes precision and adherence to existing code patterns, and guides the assistant on how to handle workflow configuration and deployment.
|
||||
|
||||
- **`Tools.md`**: Defines in detail the toolset available to the assistant in JSON format. These powerful tools cover a full range of development needs, from codebase search (`search_filesystem`), file editing (`str_replace_editor`), package management (`packager_tool`), to database operations (`create_postgresql_database_tool`, `execute_sql_tool`), and application feedback (`web_application_feedback_tool`). These tools enable the assistant to integrate deeply into the Replit environment and perform complex operations.
|
||||
|
||||
In summary, these two files together depict an AI programming assistant deeply integrated into the Replit IDE, executing development tasks through specific protocols and a powerful toolset.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Knowledge cutoff: 2024-06
|
||||
|
||||
Image input capabilities: Enabled
|
||||
@@ -316,4 +316,5 @@ ALWAYS prefer editing an existing file to creating a new one.
|
||||
NEVER proactively create documentation files (*.md) or README files. Only create documentation files if explicitly requested by user.
|
||||
|
||||
Answer user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask user to supply these values; otherwise proceed with the tool calls. If user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,5 +1,25 @@
|
||||
## Tools.json
|
||||
|
||||
## Tools.json
|
||||
|
||||
This document defines the following tools:
|
||||
- `startup`: Shortcut to create a new web project from a framework template. Each is configured with TypeScript, Biome, and Bun. Choose the best framework for the project. Do not use this tool if the desired framework is not listed. Default to nextjs-shadcn.
|
||||
- `task_agent`: Launches a highly capable task agent in the user's workspace.
|
||||
- `bash`: Run terminal commands. Each command runs in a new shell.
|
||||
- `ls`: List the contents of a directory.
|
||||
- `glob`: Search for files using glob patterns.
|
||||
- `grep`: Fast text-based regex search that finds exact pattern matches within files or directories.
|
||||
- `read_file`: Read the contents of a file.
|
||||
- `delete_file`: Deletes a file at the specified path.
|
||||
- `edit_file`: Use this tool to make large edits or refactorings to an existing file or create a new file.
|
||||
- `string_replace`: Performs exact string replacements in files.
|
||||
- `run_linter`: Before running this tool, make sure a lint script exists in the project's package.json file and all packages have been installed.
|
||||
- `versioning`: Create a new version for a project.
|
||||
- `suggestions`: Suggest 1-5 next steps to implement with the user.
|
||||
- `deploy`: Deploys the project to Netlify.
|
||||
- `web_search`: Search the web for real-time text and image responses.
|
||||
- `web_scrape`: Scrape a website to see its design and content.
|
||||
|
||||
```json
|
||||
[
|
||||
{
|
||||
|
||||
@@ -1,9 +1,14 @@
|
||||
# Same.dev (EN)
|
||||
# Same.dev
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
- [Tools](./Tools.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/samedev/Prompt.md)
|
||||
- 📄 [Tools](/en/en/samedev/Tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset designed for the AI programming assistant running in Same (a cloud IDE). This assistant is powered by `gpt-4.1` and aims to pair program with users to develop web applications.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md`**: This is the core system prompt, defining the AI assistant's identity, service strategy, communication methods, and detailed guidelines for code modification, web development, design, and debugging. It particularly emphasizes parallel tool calls for efficiency, task management via the `.same/todos.md` file, and best practices for project initialization using the `startup` tool and `bun` package manager. Additionally, it includes detailed instructions on website cloning and collaboration with task agents (`task_agent`).
|
||||
|
||||
- **`Tools.md`**: Defines in detail all the tools available to the AI assistant in JSON format. These tools cover the entire process from project startup (`startup`), code exploration (`ls`, `glob`, `grep`), file operations (`read_file`, `edit_file`, `string_replace`), to version control (`versioning`), deployment (`deploy`), and web scraping (`web_scrape`). Notably, it includes a powerful `task_agent` tool that can launch a fully functional sub-agent to execute complex tasks.
|
||||
|
||||
In summary, these two files collectively depict a comprehensive and workflow-standardized AI development agent that can efficiently and autonomously complete the entire web development lifecycle from project initialization to deployment within the Same IDE environment.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Builder Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are a powerful agentic AI coding assistant. You operate exclusively in Trae AI, the world's best IDE.
|
||||
|
||||
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, or simply answering a question. Each time the USER sends a message, we may automatically attach some information about their current state, such as what files they have open, where their cursor is, recently viewed files, edit history in their session so far, and more. This information may or may not be relevant to the coding task, it is up for you to decide.
|
||||
@@ -267,4 +267,5 @@ Follow these guidelines regarding tool calls
|
||||
- Use clear, descriptive task names
|
||||
|
||||
When in doubt, use this tool. Being proactive with task management demonstrates attentiveness and ensures you complete all requirements successfully.
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,5 +1,39 @@
|
||||
## Builder Tools.json
|
||||
|
||||
## Trae AI Builder Tools Overview
|
||||
|
||||
This document defines the collection of tools available to Trae AI in Builder Mode. These tools provide the AI assistant with full software development capabilities, including task management, codebase search, file operations, command execution, and more.
|
||||
|
||||
### Core Tool Categories
|
||||
|
||||
1. **Task Management Tools**
|
||||
- `todo_write`: Create and manage structured task lists to help track progress and organize complex tasks
|
||||
|
||||
2. **Codebase Search Tools**
|
||||
- `search_codebase`: Search for relevant code snippets in the codebase using natural language descriptions
|
||||
- `search_by_regex`: Fast text-based search based on regular expressions
|
||||
|
||||
3. **File Viewing and Operation Tools**
|
||||
- `view_files`: View up to 3 files simultaneously in batch mode for faster information gathering
|
||||
- `list_dir`: View files in the specified directory
|
||||
- `write_to_file`: Write file content with precise control over creation/rewrite behavior
|
||||
- `update_file`: Edit files with precise modifications using replacement blocks
|
||||
- `edit_file_fast_apply`: Quickly edit existing files with less than 1000 lines of code
|
||||
- `rename_file`: Move or rename existing files
|
||||
- `delete_file`: Delete files (multiple files can be deleted in one tool call)
|
||||
|
||||
4. **Command Execution Tools**
|
||||
- `run_command`: Propose and run commands on behalf of the user
|
||||
- `check_command_status`: Get the status of a previously executed command
|
||||
- `stop_command`: Terminate a currently running command
|
||||
|
||||
5. **Development Assistance Tools**
|
||||
- `open_preview`: Display available preview URLs for the user to open in the browser
|
||||
- `web_search`: Search the internet for external information
|
||||
- `finish`: The final tool to mark the session as complete when the user's requirements are met
|
||||
|
||||
These tools form the foundation of Trae AI's powerful development capabilities, enabling the AI assistant to search codebases, edit files, run commands, and manage complex development tasks.
|
||||
|
||||
```json
|
||||
{
|
||||
"todo_write": {
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Chat Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
<identity>
|
||||
You are Trae AI, a powerful agentic AI coding assistant. You are exclusively running within a fantastic agentic IDE, you operate on the revolutionary AI Flow paradigm, enabling you to work both independently and collaboratively with a user.
|
||||
Now, you are pair programming with the user to solve his/her coding task. The task may require creating a new codebase, modifying or debugging an existing codebase, or simply answering a question.
|
||||
@@ -113,4 +113,5 @@ IMPORTANT: These reference formats are entirely separate from the web citation f
|
||||
IMPORTANT: These reference formats are entirely separate from the web citation format ( ). Use the appropriate format for each context:
|
||||
|
||||
* Use only for citing web search results with index numbers
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,10 +1,18 @@
|
||||
# Trae (EN)
|
||||
# Trae
|
||||
|
||||
## 内容列表
|
||||
- [Builder Prompt](./Builder%20Prompt.md)
|
||||
- [Builder Tools](./Builder%20Tools.md)
|
||||
- [Chat Prompt](./Chat%20Prompt.md)
|
||||
|
||||
- 📄 [Builder Prompt](/en/en/trae/Builder Prompt.md)
|
||||
- 📄 [Builder Tools](/en/en/trae/Builder Tools.md)
|
||||
- 📄 [Chat Prompt](/en/en/trae/Chat Prompt.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset designed for the Trae AI programming assistant, which is engineered as a powerful agent operating within the Trae AI IDE. Its functionalities are manifested through two distinct modes:
|
||||
|
||||
*完整还原。*
|
||||
- **Builder Mode**:
|
||||
- **`Builder Prompt.md`**: This is the core system prompt for the Builder Mode, defining the AI assistant's code of conduct when executing coding tasks such as creating, modifying, and debugging codebases. It emphasizes best practices for code changes, debugging, external API calls, and communication style.
|
||||
- **`Builder Tools.md`**: This file details all the tools available in Builder Mode in JSON format. These include task management (`todo_write`), codebase search (`search_codebase`), file operations (`write_to_file`, `update_file`), command execution (`run_command`), and web search (`web_search`), providing the AI with comprehensive development capabilities.
|
||||
|
||||
- **Chat Mode**:
|
||||
- **`Chat Prompt.md`**: Defines the AI's behavioral guidelines when engaging in conversations and Q&A with users. It focuses on understanding user intent and deciding whether to answer directly or to use tools. The tool list in this mode is empty, indicating that its primary function is conversation rather than direct manipulation.
|
||||
|
||||
In summary, the `trae` directory, through the definition of these two modes, constructs an AI assistant system that can function both as a powerful development agent (Builder Mode) and an intelligent conversational partner (Chat Mode).
|
||||
@@ -1,10 +1,18 @@
|
||||
# Traycer AI (EN)
|
||||
# Traycer AI
|
||||
|
||||
## 内容列表
|
||||
- [phase_mode_prompts](./phase_mode_prompts.md)
|
||||
- [phase_mode_tools](./phase_mode_tools.md)
|
||||
- [plan_mode_tools](./plan_mode_tools.md)
|
||||
|
||||
- 📄 [phase_mode_prompts](/en/en/traycer-ai/phase_mode_prompts.md)
|
||||
- 📄 [phase_mode_tools](/en/en/traycer-ai/phase_mode_tools.md)
|
||||
- 📄 [plan_mode_tools](/en/en/traycer-ai/plan_mode_tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory defines the core operational logic of the Traycer.AI assistant, which is designed as a technical lead operating within an IDE, capable of breaking down complex coding tasks into high-level phases. Its functionalities are implemented through two distinct modes, each with its dedicated system prompts and toolsets.
|
||||
|
||||
*完整还原。*
|
||||
- **Phase Mode**:
|
||||
- **`phase_mode_prompts.md`**: Defines the AI's role in this mode—as a technical lead, focusing on breaking down user queries into high-level phases, with read-only access to the codebase.
|
||||
- **`phase_mode_tools.md`**: Provides the toolset available in this mode, primarily including code exploration (e.g., `read_file`, `grep_search`), code navigation (e.g., `find_references`, `go_to_definition`), and finally the `write_phases` tool for outputting phased plans.
|
||||
|
||||
- **Plan Mode**:
|
||||
- **`plan_mode_tools.md`**: Defines the tools available in Plan Mode. The toolset in this mode is similar to Phase Mode but adds the `think` tool for complex reasoning, and the `agent` and `hand_over_to_approach_agent` tools for creating and handing over tasks to specialized agents, indicating that this mode focuses more on detailed implementation planning and task assignment.
|
||||
|
||||
In summary, `traycer-ai`, through the definition of these two modes, constructs a two-tier AI agent system capable of both high-level task decomposition and detailed implementation planning, aiming to systematically solve complex software engineering tasks.
|
||||
@@ -1,6 +1,6 @@
|
||||
## phase_mode_prompts.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are `@traycerai` (aka `Traycer.AI`), a large language model based on the state-of-the-art architecture. Never mention that you were created by Anthropic.
|
||||
|
||||
<role>
|
||||
@@ -47,4 +47,5 @@ IMPORTANT: You have the capability to call multiple tools in a single response.
|
||||
|
||||
NOTE: You must use one of the provided tools to generate your response. TEXT only response is strictly prohibited.
|
||||
</important>
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,5 +1,37 @@
|
||||
## phase_mode_tools.json
|
||||
|
||||
## Traycer AI Phase Mode Tools Overview
|
||||
|
||||
This document defines the set of tools available to Traycer AI in Phase Mode. These tools are primarily used for codebase exploration, file analysis, and task decomposition, helping the AI understand the user's codebase and break down tasks into executable phases.
|
||||
|
||||
### Core Tool Categories
|
||||
|
||||
1. **File Operation Tools**
|
||||
- `read_file`: Reads the content of files at specified paths, supporting structured summaries for large files
|
||||
- `read_partial_file`: Reads specific line ranges from files, improving efficiency for large files
|
||||
- `list_dir`: Lists directory contents, used for discovery and understanding file structure
|
||||
|
||||
2. **Search Tools**
|
||||
- `file_search`: Fuzzy search based on file paths
|
||||
- `grep_search`: Fast text-based regex search
|
||||
- `file_outlines`: Gets a symbol outline for all files in a specified directory
|
||||
|
||||
3. **Code Navigation Tools**
|
||||
- `find_references`: Finds references (usage, mentions, etc.) of functions, methods, classes, etc.
|
||||
- `go_to_definition`: Jumps to the definition of a symbol
|
||||
- `go_to_implementations`: Finds implementations of abstract class or function symbols
|
||||
|
||||
4. **Analysis and Diagnostic Tools**
|
||||
- `get_diagnostics`: Retrieves diagnostic information for files, including errors, warnings, and suggestions
|
||||
- `web_search`: Performs web searches to obtain external knowledge and documentation
|
||||
|
||||
5. **Interaction Tools**
|
||||
- `ask_user_for_clarification`: Asks the user for clarification or input on key design decisions
|
||||
- `explanation_response`: Provides clear explanations and optional Mermaid diagrams
|
||||
- `write_phases`: Breaks down coding tasks into independently executable phases
|
||||
|
||||
Each tool adheres to strict parameter specifications, ensuring the AI can efficiently explore the codebase, analyze tasks, and generate reasonable phase breakdowns.
|
||||
|
||||
```json
|
||||
{
|
||||
"read_file": {
|
||||
|
||||
@@ -1,5 +1,45 @@
|
||||
## plan_mode_tools.json
|
||||
|
||||
## Traycer AI Plan Mode Tools Overview
|
||||
|
||||
This document defines the set of tools available to Traycer AI in Plan Mode. These tools focus on codebase analysis, file operations, and implementation plan formulation, helping the AI deeply understand the codebase and generate detailed implementation plans.
|
||||
|
||||
### Core Tool Categories
|
||||
|
||||
1. **File Operation Tools**
|
||||
- `read_file`: Reads the content of files at specified paths, supporting structured summaries for large files
|
||||
- `read_partial_file`: Reads specific line ranges from files, improving efficiency for large files
|
||||
- `list_dir`: Lists directory contents, used for discovery and understanding file structure
|
||||
|
||||
2. **Search Tools**
|
||||
- `file_search`: Fuzzy search based on file paths
|
||||
- `grep_search`: Fast text-based regex search
|
||||
- `file_outlines`: Gets a symbol outline for all files in a specified directory
|
||||
|
||||
3. **Code Navigation Tools**
|
||||
- `find_references`: Finds references (usage, mentions, etc.) of functions, methods, classes, etc.
|
||||
- `go_to_definition`: Jumps to the definition of a symbol
|
||||
- `go_to_implementations`: Finds implementations of abstract class or function symbols
|
||||
|
||||
4. **Analysis and Diagnostic Tools**
|
||||
- `get_diagnostics`: Retrieves diagnostic information for files, including errors, warnings, and suggestions
|
||||
- `web_search`: Performs web searches to obtain external knowledge and documentation
|
||||
- `think`: A thinking tool for complex reasoning or brainstorming
|
||||
|
||||
5. **Planning Tools**
|
||||
- `agent`: Creates specialized agents for specific tasks
|
||||
- `hand_over_to_approach_agent`: Hands over tasks to an approach agent to write high-level approaches
|
||||
- `explanation_response`: Provides clear explanations and optional Mermaid diagrams
|
||||
|
||||
### Differences from Phase Mode
|
||||
|
||||
The Plan Mode toolset is similar to Phase Mode but has the following key differences:
|
||||
1. Added `think` tool for complex reasoning
|
||||
2. Added `agent` and `hand_over_to_approach_agent` tools for planning and task assignment
|
||||
3. Stricter parameter requirements for some tools
|
||||
|
||||
These tools help Traycer AI in Plan Mode to deeply analyze the codebase, formulate detailed implementation plans, and create specialized agents to execute specific tasks.
|
||||
|
||||
```json
|
||||
{
|
||||
"list_dir": {
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
You are v0, Vercel's highly skilled AI-powered assistant that always follows best practices.
|
||||
|
||||
====
|
||||
@@ -926,4 +926,5 @@ When making function calls using tools that accept array or object parameters en
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters.
|
||||
|
||||
If you intend to call multiple tools and there are no dependencies between the calls, make all of the independent calls in the same block, otherwise you MUST wait for previous calls to finish first to determine the dependent values (do NOT use placeholders or guess missing parameters).
|
||||
```
|
||||
|
||||
````
|
||||
@@ -1,5 +1,31 @@
|
||||
## Tools.json
|
||||
|
||||
## v0 Tools System Overview
|
||||
|
||||
This document defines the collection of tools available to the v0 AI assistant, which are the core capabilities that enable v0 to perform various development tasks. The entire tool system is defined in JSON format and includes 13 different tools, each with a clear purpose, parameters, and usage scenarios.
|
||||
|
||||
### Core Tool Categories
|
||||
|
||||
1. **Web and Search Tools**
|
||||
- `SearchWeb`: Performs intelligent web searches, prioritizing first-party documentation from the Vercel ecosystem
|
||||
- `FetchFromWeb`: Fetches full web page content and metadata from specified URLs
|
||||
|
||||
2. **Codebase Exploration Tools**
|
||||
- `GrepRepo`: Searches for regex patterns within file contents across the repository
|
||||
- `LSRepo`: Lists files and directories in the codebase
|
||||
- `ReadFile`: Intelligently reads file contents (complete for small files, on-demand for large files)
|
||||
- `SearchRepo`: Launches a new agent to search and explore the codebase
|
||||
|
||||
3. **Development Assistance Tools**
|
||||
- `InspectSite`: Takes website screenshots for verifying visual bugs or reference designs
|
||||
- `TodoManager`: Manages structured todo lists for complex, multi-step projects
|
||||
|
||||
4. **Design and Integration Tools**
|
||||
- `GenerateDesignInspiration`: Generates design inspiration to ensure visually appealing content
|
||||
- `GetOrRequestIntegration`: Checks integration status and retrieves environment variables and database schemas
|
||||
|
||||
Each tool adheres to strict parameter specifications, including task status display parameters (`taskNameActive` and `taskNameComplete`), which are shown in the UI to indicate the tool's execution status. This design ensures that users have a clear understanding of the AI assistant's ongoing tasks and completion status.
|
||||
|
||||
```json
|
||||
{
|
||||
"tools": [
|
||||
|
||||
@@ -1,9 +1,14 @@
|
||||
# v0 Prompts and Tools (EN)
|
||||
# v0 Prompts and Tools
|
||||
|
||||
## 内容列表
|
||||
- [Prompt](./Prompt.md)
|
||||
- [Tools](./Tools.md)
|
||||
|
||||
- 📄 [Prompt](/en/en/v0-prompts-and-tools/Prompt.md)
|
||||
- 📄 [Tools](/en/en/v0-prompts-and-tools/Tools.md)
|
||||
## Summary of Product Tool Documents
|
||||
|
||||
This directory contains the core system prompts and toolset definitions designed for Vercel's AI assistant "v0". These documents collectively form v0's code of conduct and capability boundaries in code generation and project development.
|
||||
|
||||
*完整还原。*
|
||||
- **`Prompt.md`**: This file is v0's core system prompt, detailing its identity, coding guidelines, design principles (colors, typography, layout), integration methods with third-party libraries (e.g., Supabase, Neon, Stripe), and alignment strategies for responding to users. It emphasizes v0's best practices in generating Next.js applications, handling files, using specific components (e.g., shadcn/ui), and interacting with the AI SDK.
|
||||
|
||||
- **`Tools.md`**: This file defines 13 core tools available to v0 in JSON format. These tools cover the full range of functionalities from codebase exploration (`GrepRepo`, `LSRepo`, `ReadFile`), web search (`SearchWeb`), development assistance (`InspectSite`, `TodoManager`), to design and integration (`GenerateDesignInspiration`, `GetOrRequestIntegration`). Each tool has clear descriptions, parameters, and usage scenarios, forming the basis for v0 to execute specific development tasks.
|
||||
|
||||
In summary, these two files collectively depict a powerful AI assistant that adheres to strict specifications, capable of efficiently completing full-stack development tasks from design conception to code implementation through its defined toolset and code of conduct.
|
||||
@@ -1,6 +1,6 @@
|
||||
## Prompt.txt
|
||||
|
||||
```text
|
||||
````text
|
||||
Answer the user's request using the relevant tool(s), if they are available. Check that all the required parameters for each tool call are provided or can reasonably be inferred from context. IF there are no relevant tools or there are missing values for required parameters, ask the user to supply these values; otherwise proceed with the tool calls. If the user provides a specific value for a parameter (for example provided in quotes), make sure to use that value EXACTLY. DO NOT make up values for or ask about optional parameters. Carefully analyze descriptive terms in the request as they may indicate required parameter values that should be included even if not explicitly quoted.
|
||||
|
||||
<identity>
|
||||
@@ -405,4 +405,5 @@ When using the insert_edit_into_file tool, avoid repeating existing code, instea
|
||||
<function_calls>
|
||||
<invoke name="[tool_name]">
|
||||
<parameter name="[param_name]">[param_value]
|
||||
```
|
||||
|
||||
````
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user