After modifying installation scripts (install.sh, install/*.sh, or install/lib/*.sh), validate that changes don't break the installation process.

This skill is Claude-only and runs automatically before claiming work is complete.

Validation Steps

Run these checks in order. Stop if any fail and fix the issue before proceeding.

1. Shellcheck (Syntax & Best Practices)

Verify all shell scripts pass shellcheck with project standards (bash 4.0+, severity=warning):

bash

shellcheck install.sh install/**/*.sh

Must pass: No warnings or errors

2. Dry-run Installation (Minimal Profile)

Test that the installation script can parse and plan installation without making changes:

bash

./install.sh --minimal --dry-run

Expected: Clean exit (status 0), no errors in output

3. Core Test Suite

Run the BATS test suite to ensure core functionality works:

bash

make -f test.mk test

Expected: All tests pass

When to Run

Run this validation:

•After editing any file in install/ directory
•Before creating commits with installation changes
•Before claiming "installation script fixed" or similar
•Before creating PRs that modify install scripts

If Validation Fails

Shellcheck failures

•Fix syntax errors immediately
•Address warnings about quoting, logic errors
•Check .shellcheckrc for project standards

Dry-run failures

•Check error messages for clues
•
Common issues:
- •Missing function definitions
- •Incorrect sourcing of library files
- •Missing command checks (ensure commands exist before use)

Test failures

•Read failing test output carefully
•Fix the underlying issue, not the test (unless test is wrong)
•Run single test file for faster iteration: bats install/tests/test-<name>.bats

Notes

•DO NOT skip validation to save time - broken install scripts affect all users
•DO NOT modify tests to pass without fixing the actual issue
•DO fix issues incrementally (shellcheck → dry-run → tests)
•The --dry-run flag prevents actual installation, safe to run anywhere