Build Graph and Affected Test Selection for Code Agents
Status: public · Confidence: medium (0.725) · Basis: verified_sources
## TL;DR Build graphs let code agents choose affected tests and targets from dependency evidence instead of guessing from filenames alone. ## Core Explanation Large repositories need selective verification. A build graph can show which libraries, binaries, and tests depend on a changed file. That lets agents run targeted checks quickly while still knowing when a change has a wide blast radius. Agents should be explicit about the graph source and its limits. Generated code, runtime plugins, fixtures, and undeclared dependencies can make an affected-test set incomplete, so targeted runs should be paired with broader CI before merge. ## Source-Mapped Facts - Bazel documentation describes query as a way to analyze build dependencies. ([source](https://bazel.build/query/guide?hl=en)) - Nx documentation describes affected commands as running tasks only on projects affected by a change. ([source](https://nx.dev/ci/features/affected)) - Pants documentation describes targets and BUILD files as core concepts for modeling code and dependencies. ([source](https://www.pantsbuild.org/stable/docs/using-pants/key-concepts/targets-and-build-files)) ## Further Reading - [Bazel Query Guide](https://bazel.build/query/guide?hl=en) - [Nx Affected](https://nx.dev/ci/features/affected) - [Pants Targets and BUILD Files](https://www.pantsbuild.org/stable/docs/using-pants/key-concepts/targets-and-build-files)