FAQ - Frequently Asked Questions

A: While 'AI Agent' is the general term, each platform uses specific tools. **Cursor** uses 'Rules' (defined in .cursorrules files) to guide the AI. **Claude** utilizes 'Plugins' and 'Skills' to extend capabilities. **Windsurf** relies on 'Flows' and 'Cascade' context. **Replit** offers 'Templates' for quickstarts, and **MCP** (Model Context Protocol) connects 'Servers' to provide data tools. AgentDepot curates over 120 verified resources across all these platforms.

A: Cursor Rules are text-based instructions that customize your AI interactions. Browse our Cursor collection, find a rule you like (e.g., 'React Best Practices'), and click 'Copy Install'. Then, in Cursor, go to Settings > Rules > Add Rule (or create a `.cursorrules` file in your root directory) and paste the content. These rules are essentially prompt instructions that guide the AI's behavior within your project context.

A: Windsurf Cascade is an AI flow state that maintains context across your entire project. Unlike standard chat, Cascade understands your file structure and can perform complex, multi-step refactoring. You can install Windsurf-specific flows or prompts from AgentDepot to optimize Cascade for tasks like debugging or feature implementation.

A: Yes, most content on AgentDepot—including Cursor Rules, Replit Templates, and MCP Servers—is free and open-source. However, using the underlying AI models (like GPT-4 or Claude 3.5) within these tools usually requires a subscription to the respective platform (Cursor Pro, Windsurf, or Claude Pro). We clearly indicate if a specific tool requires a paid API key or subscription.

A: Standard AI chat gives generic answers. Cursor Rules inject specific context and guidelines into every interaction. For example, a 'Next.js Rule' ensures the AI only suggests App Router code, preventing outdated Page Router suggestions. This makes Cursor significantly more accurate for your specific codebase.

A: Claude Plugins (often used with Claude Code CLI) and Skills allow the AI to perform actions beyond text generation, like running terminal commands, querying databases, or managing files. They turn Claude from a chatbot into a capable coding assistant. AgentDepot lists popular plugins for testing, database management, and code review.

A: The Model Context Protocol (MCP) connects AI assistants to your data. An 'MCP Server' is a bridge that lets tools like Claude Desktop talk to external systems like GitHub, Slack, or PostgreSQL. Installing an MCP Server gives your AI direct access to those resources.

A: Replit Templates are full development environments. On AgentDepot, find a template (e.g., 'Python Flask Starter') and click the link to open it in Replit. You can then 'Fork' the template to create your own copy with all dependencies pre-installed.

A: Yes! You can combine multiple Cursor Rules (e.g., one for TypeScript and one for testing). Similarly, you can enable multiple Claude Plugins or MCP Servers. Just ensure they don't have conflicting instructions—start with a focused set and expand as needed (e.g., a linter rule and a documentation plugin).

A: Simply create a `.cursorrules` file in the root of your project. Write detailed instructions in plain English (e.g., 'Always use a specific naming convention', 'Prefer functional components'). You can refer to examples on AgentDepot for structure and formatting tips.

A: We welcome contributions! Whether you've written a great Cursor Rule, built a Replit Template, or developed an MCP Server, you can submit it via our 'Submit' page. We review all submissions to ensure they are safe and useful for the community. For verified status, ensure you have a public repository and clear documentation.

A: For **Cursor**, try the 'React Expert' rule to help with frontend code. For **Replit**, grab a 'Node.js Starter' template. For **Claude**, the 'Code Explainer' plugin is great for learning. Start small—installing too many extensions at once can be overwhelming.

A: Most AI features (Cursor's generation, Claude's answers) require internet access to reach the LLMs. However, local rules and linting configurations work offline. Some local LLMs (via tools like Ollama) can be connected to these platforms for offline use if supported. AgentDepot tags agents that support offline functionality.

A: Updates depend on the installation method. For npm-based tools, run `npm update -g <package-name>`. For Cursor Rules, you simply update the text in your `.cursorrules` file. AgentDepot lists version information so you can check if a newer version of a tool is available.

A: First, verify your `.cursorrules` file is in the root directory of your project. Check for syntax errors—rules should be plain text instructions, not code. Restart Cursor after adding the rule. If issues persist, try with a simple rule first (e.g., 'Always use TypeScript') to confirm rules are being read. Make sure you're using the latest version of Cursor.

A: Common fixes: 1) Verify the server is installed correctly (`npm list -g <server-name>`). 2) Check your Claude Desktop configuration file (usually at `~/Library/Application Support/Claude/` on Mac). 3) Restart Claude Desktop after configuration changes. 4) Check the server's GitHub repository for specific setup requirements. 5) Look for port conflicts if the server uses a specific port.

A: Use our search bar with specific keywords (e.g., 'react typescript' instead of just 'react'). Use the tool filters (Cursor, Windsurf, etc.) to narrow results. Try the category filters (Web, Coding, Data). Check the tags on similar agents for alternative search terms. If you still can't find it, email us at hello@agentdepot.dev with details about what you're looking for.

A: Email us immediately at hello@agentdepot.dev with: 1) The agent name and URL, 2) Description of the issue, 3) Steps to reproduce (if applicable), 4) Impact/severity. For security issues, we respond within 48 hours and remove problematic agents immediately. You can also report via the agent's GitHub repository if available.

A: Visit our Submit page (/submit) and choose one of three methods: 1) **GitHub PR** (recommended): Fork our agentdepot-agents repository, add your agent to the appropriate file (cursor.ts, mcp.ts, etc.), and submit a PR. 2) **GitHub Issue**: Open an 'Agent Submission' issue with details. 3) **Email**: Send agent details to hello@agentdepot.dev. Include: agent name, description, tool type, installation instructions, GitHub URL (if applicable), and your contact info.

A: We review all submissions within 24-48 hours. Requirements: 1) **Works correctly** - we test each agent, 2) **Clear documentation** - installation instructions must be detailed, 3) **Safe** - no malicious code or obfuscated scripts, 4) **Original or properly attributed** - respect licenses and give credit, 5) **Relevant** - must be for AI coding tools (Cursor, Windsurf, Claude, MCP, Replit). High-quality submissions with public GitHub repos are prioritized.

A: Yes! For updates: submit a new PR to the agentdepot-agents repository or email us at hello@agentdepot.dev with the changes. For removal: email us with the agent name and reason for removal. We'll process updates within 24-48 hours. You retain all rights to your submissions.

A: Email us at hello@agentdepot.dev for all inquiries: agent submissions, bug reports, feature requests, partnership opportunities, or general questions. We respond within 24-48 hours during business days. For urgent security issues, mark your email subject with [URGENT].

A: If you believe content on AgentDepot infringes your copyright, send a DMCA takedown notice to hello@agentdepot.dev. Include: 1) Identification of your copyrighted work, 2) URL of the infringing content, 3) Your contact information, 4) A statement that you have a good faith belief the use is unauthorized, 5) Your physical or electronic signature. We respond to valid DMCA notices within 48-72 hours.

A: We're currently focused on building the core directory. A Discord community or forum may be launched in the future based on user demand. For now, connect with us via email (hello@agentdepot.dev) or through our GitHub repository for discussions and contributions.