Enhance AGENTS.md with bash patterns and project learnings from PR #28

[Copilot]

The AGENTS.md file lacked critical bash-specific patterns and project context needed for effective AI agent interaction, particularly lessons from PR #28's exit code handling implementation.

Enhanced Documentation Structure

Bash-Specific Patterns (new section)

• Exit code handling: SCRIPT_DIALOG_CANCEL_EXIT_CODE (124), command substitution propagation pattern || exit "$?", and exit status capture timing semantics
• Variable existence checking: ${VAR+x} pattern
• Platform detection: actual implementation patterns with [[ ]] for wildcards, [ ] for POSIX tests

Cross-Platform Considerations (new section)

• Platform-specific behaviors: macOS, Windows/WSL, Linux desktop detection via pgrep -l
• Wayland vs X11 compositor detection for screenshot utility

Testing Approaches (new section)

• Manual: interface selection, cancel behavior validation
• Screenshot: screenshot-dialogs.sh usage patterns
• CI: shellcheck automation

Common Patterns & Pitfalls (new section)

• DO: Capture $? immediately after commands, test multiple interfaces
• DON'T: Capture $? after if/fi (always returns 0), break cancel detection

PR #28 Learnings (new section)

• Refactoring trade-offs, cancellation testing importance, exit code 124 rationale

Critical Pattern Example

# CORRECT - exit status captured immediately
if [ "$INTERFACE" == "zenity" ]; then
    result=$(zenity --entry --text="$1" 2>/dev/null)
    exit_status=$?  # Must be right after the command
fi

# WRONG - captures if statement exit status (always 0)
if [ "$INTERFACE" == "zenity" ]; then
    result=$(zenity --entry --text="$1" 2>/dev/null)
fi
exit_status=$?  # This is 0, not the zenity exit code

Stats

• 53 → 214 lines (+172, -10)
• Added 5 major sections
• Enhanced all existing sections with project-specific patterns

Original prompt

Are there any improvements to the AGENTS.md that can be made based on the project and feedback given on all the PRs?

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.