diff --git a/src/dda/tools/go.py b/src/dda/tools/go.py
index 311de4a8..168ee24d 100644
--- a/src/dda/tools/go.py
+++ b/src/dda/tools/go.py
@@ -11,7 +11,9 @@
 from dda.utils.fs import Path
 
 if TYPE_CHECKING:
-    from collections.abc import Generator
+    from collections.abc import Generator, Iterable
+    from os import PathLike
+    from typing import Any
 
 
 class Go(Tool):
@@ -62,3 +64,68 @@ def version(self) -> str | None:
             return match.group(1)
 
         return None
+
+    def _build(self, args: list[str], **kwargs: Any) -> str:
+        """Run a raw go build command."""
+        return self.capture(["build", *args], check=True, **kwargs)
+
+    def build(
+        self,
+        *packages: str | PathLike,
+        output: str | PathLike,
+        build_tags: set[str] | None = None,
+        gcflags: Iterable[str] | None = None,
+        ldflags: Iterable[str] | None = None,
+        env_vars: dict[str, str] | None = None,
+        force_rebuild: bool = False,
+        **kwargs: Any,
+    ) -> str:
+        """
+        Run an instrumented Go build command.
+
+        Args:
+            packages: The go packages to build, passed as a list of strings or Paths.
+                Empty by default, which is equivalent to building the current directory.
+            output: The path to the output binary.
+            build_tags: Build tags to include when compiling. Empty by default.
+            gcflags: The gcflags (go compiler flags) to use, passed as a list of strings. Empty by default.
+            ldflags: The ldflags (go linker flags) to use, passed as a list of strings. Empty by default.
+            env_vars: Extra environment variables to set for the build command. Empty by default.
+            force_rebuild: Whether to force a rebuild of the package and bypass the build cache.
+            **kwargs: Additional arguments to pass to the go build command.
+        """
+        from platform import machine as architecture
+
+        from dda.config.constants import Verbosity
+        from dda.utils.platform import PLATFORM_ID
+
+        command_parts = [
+            "-trimpath",  # Always use trimmed paths instead of absolute file system paths # NOTE: This might not work with delve
+            "-mod=readonly",  # Always use readonly mode, we never use anything else
+            f"-o={output}",
+        ]
+
+        if force_rebuild:
+            command_parts.append("-a")
+
+        # Enable data race detection on platforms that support it (all except windows arm64)
+        if not (PLATFORM_ID == "windows" and architecture() == "arm64"):
+            command_parts.append("-race")
+
+        if self.app.config.terminal.verbosity >= Verbosity.VERBOSE:
+            command_parts.append("-v")
+        if self.app.config.terminal.verbosity >= Verbosity.DEBUG:
+            command_parts.append("-x")
+
+        if gcflags:
+            command_parts.append(f"-gcflags={' '.join(gcflags)}")
+        if ldflags:
+            command_parts.append(f"-ldflags={' '.join(ldflags)}")
+
+        if build_tags:
+            command_parts.extend(("-tags", f"{','.join(sorted(build_tags))}"))
+
+        command_parts.extend(str(package) for package in packages)
+
+        # TODO: Debug log the command parts ?
+        return self._build(command_parts, env=env_vars, **kwargs)
diff --git a/tests/conftest.py b/tests/conftest.py
index 6864cc57..0ac1920b 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -232,21 +232,37 @@ def pytest_runtest_setup(item):
         if marker.name == "requires_windows" and PLATFORM_ID != "windows":
             pytest.skip("Not running on Windows")
 
+        if marker.name == "skip_windows" and PLATFORM_ID == "windows":
+            pytest.skip("Test should be skipped on Windows")
+
         if marker.name == "requires_macos" and PLATFORM_ID != "macos":
             pytest.skip("Not running on macOS")
 
+        if marker.name == "skip_macos" and PLATFORM_ID == "macos":
+            pytest.skip("Test should be skipped on macOS")
+
         if marker.name == "requires_linux" and PLATFORM_ID != "linux":
             pytest.skip("Not running on Linux")
 
-        if marker.name == "requires_unix" and PLATFORM_ID == "windows":
-            pytest.skip("Not running on a Linux-based platform")
+        if marker.name == "skip_linux" and PLATFORM_ID == "linux":
+            pytest.skip("Test should be skipped on Linux")
+
+        if marker.name == "requires_unix" and PLATFORM_ID not in {"linux", "macos"}:
+            pytest.skip("Not running on a Unix-based platform")
+
+        if marker.name == "skip_unix" and PLATFORM_ID in {"linux", "macos"}:
+            pytest.skip("Test should be skipped on Unix-based platforms")
 
 
 def pytest_configure(config):
     config.addinivalue_line("markers", "requires_ci: Tests intended for CI environments")
     config.addinivalue_line("markers", "requires_windows: Tests intended for Windows operating systems")
+    config.addinivalue_line("markers", "skip_windows: Tests should be skipped on Windows operating systems")
     config.addinivalue_line("markers", "requires_macos: Tests intended for macOS operating systems")
+    config.addinivalue_line("markers", "skip_macos: Tests should be skipped on macOS operating systems")
     config.addinivalue_line("markers", "requires_linux: Tests intended for Linux operating systems")
+    config.addinivalue_line("markers", "skip_linux: Tests should be skipped on Linux operating systems")
     config.addinivalue_line("markers", "requires_unix: Tests intended for Linux-based operating systems")
+    config.addinivalue_line("markers", "skip_unix: Tests should be skipped on Unix-based operating systems")
 
     config.getini("norecursedirs").remove("build")  # /tests/cli/build
diff --git a/tests/tools/go/__init__.py b/tests/tools/go/__init__.py
new file mode 100644
index 00000000..79ca6026
--- /dev/null
+++ b/tests/tools/go/__init__.py
@@ -0,0 +1,3 @@
+# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <dev@datadoghq.com>
+#
+# SPDX-License-Identifier: MIT
diff --git a/tests/tools/go/conftest.py b/tests/tools/go/conftest.py
new file mode 100644
index 00000000..4ffd83f5
--- /dev/null
+++ b/tests/tools/go/conftest.py
@@ -0,0 +1,12 @@
+from base64 import urlsafe_b64encode
+from os import urandom
+
+import pytest
+
+
+@pytest.fixture
+def get_random_filename():
+    def _get_random_filename(k: int = 10) -> str:
+        return urlsafe_b64encode(urandom(k)).decode("utf-8")
+
+    return _get_random_filename
diff --git a/tests/tools/go/fixtures/small_go_project/debug.go b/tests/tools/go/fixtures/small_go_project/debug.go
new file mode 100644
index 00000000..5fefc6ee
--- /dev/null
+++ b/tests/tools/go/fixtures/small_go_project/debug.go
@@ -0,0 +1,9 @@
+//go:build debug
+// +build debug
+
+package main
+
+// Also defined in prod.go to test build tags
+func TagInfo() string {
+	return "DEBUG"
+}
diff --git a/tests/tools/go/fixtures/small_go_project/go.mod b/tests/tools/go/fixtures/small_go_project/go.mod
new file mode 100644
index 00000000..94195dd5
--- /dev/null
+++ b/tests/tools/go/fixtures/small_go_project/go.mod
@@ -0,0 +1,3 @@
+module github.com/DataDog/datadog-agent-dev/tests/tools/go/fixtures/small_go_project
+
+go 1.21.0
diff --git a/tests/tools/go/fixtures/small_go_project/main.go b/tests/tools/go/fixtures/small_go_project/main.go
new file mode 100644
index 00000000..df5140a4
--- /dev/null
+++ b/tests/tools/go/fixtures/small_go_project/main.go
@@ -0,0 +1,21 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"runtime"
+)
+
+func main() {
+	fmt.Printf("Hello from small go project!\n")
+	fmt.Printf("Go version: %s\n", runtime.Version())
+	fmt.Printf("OS/Arch: %s/%s\n", runtime.GOOS, runtime.GOARCH)
+
+	// Test command line args
+	if len(os.Args) > 1 {
+		fmt.Printf("Arguments: %v\n", os.Args[1:])
+	}
+
+	// Call build-specific function to test build tags
+	fmt.Printf("Tag: %s\n", TagInfo())
+}
diff --git a/tests/tools/go/fixtures/small_go_project/prod.go b/tests/tools/go/fixtures/small_go_project/prod.go
new file mode 100644
index 00000000..983d3457
--- /dev/null
+++ b/tests/tools/go/fixtures/small_go_project/prod.go
@@ -0,0 +1,9 @@
+//go:build !debug
+// +build !debug
+
+package main
+
+// Also defined in debug.go to test build tags
+func TagInfo() string {
+	return "PRODUCTION"
+}
diff --git a/tests/tools/go/test_go.py b/tests/tools/go/test_go.py
new file mode 100644
index 00000000..fa643818
--- /dev/null
+++ b/tests/tools/go/test_go.py
@@ -0,0 +1,113 @@
+# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <dev@datadoghq.com>
+#
+# SPDX-License-Identifier: MIT
+from __future__ import annotations
+
+import platform
+
+import pytest
+
+from dda.utils.fs import Path
+
+
+def test_default(app):
+    with app.tools.go.execution_context([]) as context:
+        assert context.env_vars == {}
+
+
+class TestPrecedence:
+    def test_workspace_file(self, app, temp_dir):
+        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
+        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
+            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Z"}
+
+    def test_module_file(self, app, temp_dir):
+        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
+        (temp_dir / "go.mod").write_text("stuff\ngo X.Y.Zrc1\nstuff")
+        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
+            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Zrc1"}
+
+    def test_version_file(self, app, temp_dir):
+        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
+        (temp_dir / "go.mod").write_text("stuff\ngo X.Y.Zrc1\nstuff")
+        (temp_dir / ".go-version").write_text("X.Y.Zrc2")
+        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
+            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Zrc2"}
+
+
+class TestBuild:
+    @pytest.mark.parametrize(
+        "call_args",
+        [
+            {"n_packages": 2},
+            {"n_packages": 1, "build_tags": {"debug"}},
+            {"n_packages": 0, "build_tags": {"prod"}, "gcflags": ["all=-N -l"], "ldflags": ["all=-s -w", "-dumpdep"]},
+            {"n_packages": 2, "build_tags": {"prod"}, "gcflags": ["all=-N -l"], "ldflags": ["all=-s -w", "-dumpdep"]},
+            {"n_packages": 0, "force_rebuild": True},
+        ],
+    )
+    def test_command_formation(self, app, mocker, call_args, get_random_filename):
+        # Patch the raw _build method to avoid running anything
+        mocker.patch("dda.tools.go.Go._build", return_value="output")
+
+        # Generate dummy package and output paths
+        n_packages = call_args.pop("n_packages", 0)
+        packages: tuple[str, ...] = tuple(get_random_filename() for _ in range(n_packages))
+        output: Path = get_random_filename()
+        app.tools.go.build(
+            *packages,
+            output=output,
+            **call_args,
+        )
+
+        flags = {
+            ("-trimpath",),
+            ("-mod=readonly",),
+            (f"-o={output}",),
+            # ("-v",), # By default verbosity is INFO
+            # ("-x",),
+        }
+
+        if not (platform.system() == "Windows" and platform.machine() == "arm64"):
+            flags.add(("-race",))
+
+        if call_args.get("build_tags"):
+            flags.add(("-tags", ",".join(sorted(call_args.get("build_tags", [])))))
+        if call_args.get("gcflags"):
+            flags.add((f"-gcflags={' '.join(call_args.get('gcflags'))}",))
+        if call_args.get("ldflags"):
+            flags.add((f"-ldflags={' '.join(call_args.get('ldflags'))}",))
+        if call_args.get("force_rebuild"):
+            flags.add(("-a",))
+
+        seen_command_parts = app.tools.go._build.call_args[0][0]  # noqa: SLF001
+
+        flags_len = len(flags)
+        seen_flags: list[str] = seen_command_parts[: flags_len + 1]
+        for flag_tuple in flags:
+            assert flag_tuple[0] in seen_flags
+
+            if len(flag_tuple) > 1:
+                flag_index = seen_flags.index(flag_tuple[0])
+                assert seen_flags[flag_index + 1] == flag_tuple[1]
+
+        if n_packages > 0:
+            assert seen_command_parts[-len(packages) :] == [str(package) for package in packages]
+
+    # This test is quite slow, we'll only run it in CI
+    @pytest.mark.requires_ci
+    @pytest.mark.skip_macos  # Go binary is not installed on macOS CI runners
+    def test_build_project(self, app, temp_dir):
+        for tag, output_mark in [("prod", "PRODUCTION"), ("debug", "DEBUG")]:
+            with (Path(__file__).parent / "fixtures" / "small_go_project").as_cwd():
+                app.tools.go.build(
+                    ".",
+                    output=(temp_dir / "testbinary").absolute(),
+                    build_tags={tag},
+                    force_rebuild=True,
+                )
+
+                assert (temp_dir / "testbinary").is_file()
+                output = app.subprocess.capture(str(temp_dir / "testbinary"))
+                assert output_mark in output
+                # Note: doing both builds in the same test with the same name also allows us to test the force rebuild
diff --git a/tests/tools/test_go.py b/tests/tools/test_go.py
deleted file mode 100644
index 49d68dfb..00000000
--- a/tests/tools/test_go.py
+++ /dev/null
@@ -1,29 +0,0 @@
-# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <dev@datadoghq.com>
-#
-# SPDX-License-Identifier: MIT
-from __future__ import annotations
-
-
-def test_default(app):
-    with app.tools.go.execution_context([]) as context:
-        assert context.env_vars == {}
-
-
-class TestPrecedence:
-    def test_workspace_file(self, app, temp_dir):
-        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
-        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
-            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Z"}
-
-    def test_module_file(self, app, temp_dir):
-        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
-        (temp_dir / "go.mod").write_text("stuff\ngo X.Y.Zrc1\nstuff")
-        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
-            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Zrc1"}
-
-    def test_version_file(self, app, temp_dir):
-        (temp_dir / "go.work").write_text("stuff\ngo X.Y.Z\nstuff")
-        (temp_dir / "go.mod").write_text("stuff\ngo X.Y.Zrc1\nstuff")
-        (temp_dir / ".go-version").write_text("X.Y.Zrc2")
-        with temp_dir.as_cwd(), app.tools.go.execution_context([]) as context:
-            assert context.env_vars == {"GOTOOLCHAIN": "goX.Y.Zrc2"}