By exploiting an injection flaw in GitHub's internal protocol, any authenticated user could execute arbitrary commands on GitHub's backend servers with a single
git pushcommand - using nothing but a standard git client.
This research was made possible by AI-augmented reverse engineering tooling, particularly IDA MCP, which allowed us to rapidly analyze compiled binaries and reconstruct internal protocols at a speed that would not have been feasible manually. As these tools continue to mature, we expect them to play an increasingly important role in uncovering vulnerability classes that require deep cross-component analysis.