What problem does it solve? It ensures agent features work identically in both local and remote execution environments by exercising the app-server/exec-server split, catching bugs that only appear when code runs on a remote executor. ## Core Features & Use Cases - Remote Test Opt-In: Use TestCodexBuilder::build_with_auto_env() in core tests and TestAppServer::new_with_auto_env() in app-server tests to run against remote executors. - Two Executor Flavors: Run tests against a Linux exec-server in Docker via scripts/test-remote-env.sh, or a Windows exec-server under Wine via Bazel targets. - Granular Test Skips: Skip tests per configuration with macros like skip_if_wine_exec!, skip_if_remote!, or skip_if_no_remote_env! when behavior legitimately differs. - Use Case: After adding a feature to the app-server, run the full integration suite against a Docker remote executor to verify thread start requests and environments behave the same as local execution. ## Quick Start Run the core integration tests against a Docker remote executor by sourcing scripts/test-remote-env.sh and then running just test -p codex-core --test all from the codex-rs directory.