test/runner.test.ts (1)

644-730: ⚡ Quick win

Add explicit tests for the two newly introduced fallback branches.

Current updates validate the preferred sha256sum path, but the new shasum fallback and “neither tool available” failure branch should also be asserted to prevent regressions.

Suggested test additions

+    it("install-openshell.sh falls back to shasum when sha256sum is unavailable", () => {
+      // stub command -v sha256sum => fail, command -v shasum => success
+      // stub shasum() to log args and return success
+      // assert checksum log contains: "SHASUM -a 256 -c -"
+    });
+
+    it("install-openshell.sh fails clearly when no SHA-256 tool is available", () => {
+      // stub command -v sha256sum => fail and command -v shasum => fail
+      // run script and assert non-zero status
+      // assert output contains: "No SHA-256 tool available (sha256sum/shasum)"
+    });

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test/runner.test.ts` around lines 644 - 730, Tests only cover the primary
sha256sum path; add two new specs in test/runner.test.ts that exercise the
shasum fallback and the "neither tool available" failure: (1) create a tmpBin
stub where sha256sum is absent but shasum (or shasum -a 256) is present (write a
small shim that logs invocations to checksumLog), source install-openshell.sh
and assert output contains the shasum-fallback marker and that checksumLog
contains the expected "-c -" invocation; (2) create a tmpBin with no
sha256sum/shasum and with gh absent/failing, run the script and assert it exits
non-zero (or prints the expected failure message such as "gh CLI download
failed"), cleaning up tmpBin afterwards; use the existing scriptPath and
checksumLog variables and mirror the style of the existing tests for spawnSync
invocation and assertions.