Repo-Pilot

Modern developers often spend significant time understanding unfamiliar codebases before they can safely implement a new feature, resolve an issue, or modify existing functionality. They must manually trace files, dependencies, APIs, architecture, and existing tests to determine what will be affected by a change. This process is time-consuming, requires repeated context switching, and can lead to overlooked dependencies, incomplete implementations, and insufficient testing. Our solution, Repo-Pilot, is an AI-powered developer workflow assistant designed to reduce this friction. It combines repository intelligence with IBM Bob 2.0 to transform a developer's task or issue into an actionable understanding of the required change. The system first analyzes a repository and builds a structured representation of its technologies, architecture, dependencies, entry points, documentation, and testing structure. When a developer provides a task such as "add authentication" or "modify the certificate verification process," IBM Bob 2.0 orchestrates specialized subagents to investigate the repository in parallel. An Architecture Agent identifies relevant components and execution flows, an Impact Agent determines potentially affected files and dependencies, and a Testing Agent identifies existing coverage and recommends additional test cases. A planning agent then combines these findings into a clear implementation plan, including relevant files, potential risks, and validation steps. Developers interact with the system through a conversational interface, allowing them to describe tasks naturally and ask follow-up questions about the generated analysis. Instead of simply generating code, the system focuses on improving the complete developer workflow from understanding a task to planning and validating a change. The solution is creative because it treats AI as a coordinated development workflow rather than a single coding assistant. IBM Bob 2.0's agent mode, parallel task execution, subagents, and document understanding can be used to divide repository analysis into specialized investigations and then synthesize their results into one actionable developer plan. The primary target users are software developers, engineering teams, and contributors working on unfamiliar or complex repositories. By automating repetitive repository investigation and organizing the results into actionable insights, Repo-Pilot aims to reduce time spent searching and understanding code, minimize missed dependencies and testing gaps, and help developers move from an issue to implementation more efficiently and confidently.