Is your feature request related to a problem? Please describe.

PR #227 (Windows native support) replaced 100+ Unicode characters across 38 source files that caused charmap codec errors on Windows cp1252 terminals (emojis, box-drawing chars, em dashes). There is no Copilot instruction preventing their reintroduction, so AI-generated code will keep producing emojis and Unicode symbols. Additionally, .github/instructions/cli.instructions.md still references the old emoji-based STATUS_SYMBOLS convention even though the codebase now uses ASCII bracket notation.

Describe the solution you'd like

1. Add a cross-platform encoding rule to copilot-instructions.md — all source code and CLI output must stay within ASCII/cp1252. Use bracket notation for status symbols: [+] success, [!] warning, [x] error, [i] info, [*] action, [>] running.
2. Update .github/instructions/cli.instructions.md — replace emoji references with the ASCII bracket convention.
3. Optionally add a dedicated .github/instructions/encoding.instructions.md with applyTo: "**".

Describe alternatives you've considered

• Lint rule only (no Copilot instruction): catches violations but doesn't prevent AI from generating them.
• Allow Unicode with Rich fallback: rejected — PR feat: add phase 1 Windows native support #227 showed dual-path output is fragile across 38+ files.

Additional context

Origin: PR #227 merge comment by @danielmeppiel. The ASCII bracket notation is already established in the codebase post-merge.