I'm developing a system that uses graph differentials to understand what has changed between library versions and upgrades the target system without breaking things. [0]
Because source isn’t always available, it scans the bytecode of an application and the new library, building a full graph of each component in Neo4j to determine what breaking changes impact the target application. This is then translated into tickets and prompts to drive an LLM to make the appropriate changes.
Handling library upgrades is rarely interesting and just adds to our overall technical debt, so it has been nice to automate it away so that we can focus on features and functionality. It supports Java and .Net currently and we’re actively adding support for other languages.
Because source isn’t always available, it scans the bytecode of an application and the new library, building a full graph of each component in Neo4j to determine what breaking changes impact the target application. This is then translated into tickets and prompts to drive an LLM to make the appropriate changes.
Handling library upgrades is rarely interesting and just adds to our overall technical debt, so it has been nice to automate it away so that we can focus on features and functionality. It supports Java and .Net currently and we’re actively adding support for other languages.
[0] https://codelogic.com/