IDD: Problem-Driven AI-Assisted Development Methodology

IDD (Issue-Driven Development) is a new development paradigm that combines issue tracking systems with AI agent routing mechanisms. Its core proposition is to make issues the central hub of development activities, with all code changes centered around clear issues. Through structured workflows and intelligent routing, it provides a reproducible work framework for the era of AI-assisted development, enhancing development traceability, measurability, and human-machine collaboration efficiency.

Traditional software development is code-centric, with issue tracking tools (such as Jira, GitHub Issues) marginalized—associations between issues and code are implicit and easily overlooked. IDD reverses this relationship, treating issues as the 'atomic units' of development. All feature implementations, bug fixes, etc., correspond to clear issues, enabling teams to trace feature motivations and decision-making processes while providing key inputs for AI-assisted programming.

IDD includes two core components: 1. issue-driven-dev: Defines the issue lifecycle (Create → Analyze → Assign → Implement → Verify → Close) based on a state machine model, with clear conditions for each state transition; 2. idd-route: A data-driven intelligent routing system that automatically assigns issues to the most suitable AI agent or human developer by analyzing issue characteristics (keywords, complexity, etc.), combining historical data and load balancing.

IDD clearly defines AI intervention scenarios: 1. Issues with clear patterns and sufficient context (e.g., generating client-side code, implementing function logic) are led by AI; 2. Creative design or complex trade-off issues (architectural design, technology selection) are led by humans, with AI assisting by providing references; 3. Sensitive/professional fields (security code, financial logic) require full human participation; idd-route is responsible for decision-making and assignment.

IDD plans a complete ecosystem: 1. idd-bench: Establishes standardized test scenarios to evaluate metrics such as routing accuracy and AI code acceptance rate; 2. idd-stats: Collects workflow data to generate visual reports and trend analysis; 3. idd-codex-companion: Bridges IDD with AI assistants in IDEs, synchronizing issue context to generate more relevant code suggestions.

IDD is compatible with mainstream toolchains: It can integrate with issue tracking systems like GitHub Issues and Jira, and work with AI assistants such as GitHub Copilot; it standardizes commit message formats and branch naming to ensure code-issue associations; it combines with existing practices like CI/CD pipelines, agile Sprint planning, and GitFlow to reduce the adoption threshold for teams.

Implementation challenges include cultural transformation (adapting to 'no code without issues'), tool integration, AI capability boundary assessment, and measurement feedback. Recommendations: Conduct small-scale pilots to accumulate cases, prioritize integration with commonly used tools, use conservative routing initially (assign uncertain issues to humans), and establish a key metrics dashboard for regular review and adjustment.

IDD is an organizational mindset for the era of AI-assisted development, emphasizing structured issue definition and human-machine collaboration. It does not replace humans but establishes a collaboration framework. As ecosystem components are implemented, it is expected to become an important part of AI-native development toolchains, providing a reference framework for teams exploring AI-assisted development.