Frequently Asked Questions about AgentPM

Q: What is AgentPM?

A: AgentPM is a planning and orchestration system designed as an MCP (Model Context Protocol) server for AI-driven software development. It helps developers plan, prioritize, and execute complex projects by acting as a virtual product manager.

Q: How does AgentPM integrate with IDEs?

A: AgentPM integrates with any IDE that supports Anthropic’s Model Context Protocol (MCP), including Cursor, Augment, VS Code Copilot, Cline, and Roo. It is installed locally and connects through the IDE’s MCP configuration.

Q: What are the prerequisites for using AgentPM?

A: The prerequisites include Node.js (version 20.0.0 or higher), an Anthropic API key for Claude AI integration, and a Perplexity API key for research-backed task generation.

Q: What are the main features of AgentPM?

A: The main features include:

• Comprehensive requirements development
• Intelligent task breakdown
• Context-aware orchestration
• Dynamic documentation generation
• Integrated documentation retrieval

Q: How does AgentPM optimize token usage?

A: AgentPM consolidates functionality around a core set of dynamic tools that are contextually economical and easy for coding agents to use and understand. It delivers the right information at the right time, optimizing token usage and eliminating the need for manual context handling.

Q: Can AgentPM work with existing project documentation?

A: Yes, AgentPM can work with or without existing documentation. It guides you through a structured interview when starting from scratch or easily adapts to an existing project or plan.

Q: What AI models does AgentPM use?

A: AgentPM primarily uses Claude Sonnet 3.7 for consistent task generation. It also supports optional Perplexity API integration for research-backed results.

Q: How does AgentPM handle changing project requirements?

A: AgentPM adaptively updates future tasks based on completed work to handle implementation drift and maintains living documentation that evolves with the project.

Q: What is the purpose of the Perplexity API key?

A: The Perplexity API key is used for integrating research-backed results into task generation, providing more accurate and comprehensive information.

Q: How do I enable debug mode in AgentPM?

A: You can enable debug mode by setting the DEBUG_LOGS environment variable to true. This will enable detailed logging to files in the logs directory.

Q: What are the different actions available in the apm_task tool?

A: The apm_task tool supports actions like get_all, get_single, get_next, filter_by_status, and filter_by_priority to query tasks in the project.

Q: How can I create a new task using AgentPM?

A: You can create a new task using the apm_task_modify tool with the create action, providing details such as title, description, priority, and dependencies.

Q: What is the purpose of the apm_project_brief_create tool?

A: The apm_project_brief_create tool creates a project brief through an interactive interview process and generates tasks based on the collected information.

Q: How can I manage task dependencies in AgentPM?

A: You can manage task dependencies using the apm_dependencies tool with actions like add, remove, validate, and fix.

Q: What does the apm_complexity tool do?

A: The apm_complexity tool analyzes task complexity, generates expansion recommendations, and creates reports to help break down complex tasks into manageable subtasks.

Q: How does AgentPM ensure code quality?

A: AgentPM incorporates software development best practices with opinionated recommendations that improve code quality and promote adherence to industry standards.

Q: Where are the task files generated by AgentPM stored?

A: Task files are generated in the apm-artifacts directory within your project, by default. This location can be customized using configuration variables.

Q: What if I encounter issues or errors while using AgentPM?

A: Check the log files generated when DEBUG_LOGS is enabled for detailed information. Also, consult the documentation for your specific IDE and the AgentPM documentation for troubleshooting tips.