Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Detect terminal capabilities including color support, TTY status, size, and Unicode support for adaptive CLI output.
.claude/skills/a5c-ai-terminal-capability-detector/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -74% | 0% |
Detect terminal capabilities for adaptive CLI output.
typescriptimport process from 'process'; import tty from 'tty'; export interface TerminalCapabilities { isTTY: boolean; colorLevel: 0 | 1 | 2 | 3; supportsUnicode: boolean; columns: number; rows: number; isCI: boolean; } export function detectCapabilities(): TerminalCapabilities { const isTTY = tty.isatty(1); const isCI = Boolean(process.env.CI || process.env.CONTINUOUS_INTEGRATION); let colorLevel: 0 | 1 | 2 | 3 = 0; if (isTTY && !process.env.NO_COLOR) { if (process.env.COLORTERM === 'truecolor') colorLevel = 3; else if (process.env.TERM?.includes('256color')) colorLevel = 2; else if (process.env.TERM && process.env.TERM !== 'dumb') colorLevel = 1; } const supportsUnicode = process.platform !== 'win32' || process.env.WT_SESSION || process.env.TERM_PROGRAM === 'vscode'; return { isTTY, colorLevel, supportsUnicode, columns: process.stdout.columns || 80, rows: process.stdout.rows || 24, isCI, }; }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,362 | 13,196 | -1% | 1 | 1 | 0% | 2,863 | 3,350 | +17% | 0 | 0 | — |
case-02 | fail→pass | 17,730 | 17,975 | +1% | 1 | 1 | 0% | 4,074 | 3,557 | -13% | 0 | 0 | — |
case-03 | fail→pass | 16,205 | 18,120 | +12% | 1 | 1 | 0% | 3,425 | 4,223 | +23% | 0 | 0 | — |
case-04 | fail→fail | 9,350 | 3,970 | -58% | 1 | 1 | 0% | 1,497 | 1,180 | -21% | 0 | 0 | — |
case-05 | fail→fail | 4,036 | 2,090 | -48% | 1 | 1 | 0% | 697 | 807 | +16% | 0 | 0 | — |
case-06 | fail→fail | 4,596 | 3,308 | -28% | 1 | 1 | 0% | 787 | 957 | +22% | 0 | 0 | — |
case-07 | fail→fail | 5,319 | 4,176 | -21% | 1 | 1 | 0% | 962 | 1,230 | +28% | 0 | 0 | — |
case-08 | fail→pass | 17,381 | 11,066 | -36% | 1 | 1 | 0% | 3,174 | 2,763 | -13% | 0 | 0 | — |
case-09 | pass→pass | 9,115 | 9,392 | +3% | 1 | 1 | 0% | 1,817 | 1,980 | +9% | 0 | 0 | — |
case-10 | fail→fail | 7,199 | 4,676 | -35% | 1 | 1 | 0% | 1,296 | 1,323 | +2% | 0 | 0 | — |
case-11 | fail→fail | 2,868 | 2,614 | -9% | 1 | 1 | 0% | 501 | 877 | +75% | 0 | 0 | — |
case-12 | fail→pass | 17,776 | 2,248 | -87% | 1 | 1 | 0% | 3,267 | 857 | -74% | 0 | 0 | — |
case-13 | fail→fail | 11,597 | 3,029 | -74% | 1 | 1 | 0% | 2,183 | 1,006 | -54% | 0 | 0 | — |
case-14 | fail→fail | 6,378 | 3,400 | -47% | 1 | 1 | 0% | 1,240 | 1,092 | -12% | 0 | 0 | — |
case-15 | fail→fail | 8,262 | 2,368 | -71% | 1 | 1 | 0% | 1,310 | 819 | -37% | 0 | 0 | — |
case-16 | fail→fail | 6,867 | 1,941 | -72% | 1 | 1 | 0% | 1,334 | 786 | -41% | 0 | 0 | — |
case-17 | fail→fail | 5,782 | 2,017 | -65% | 1 | 1 | 0% | 1,139 | 834 | -27% | 0 | 0 | — |
case-18 | fail→fail | 12,616 | 5,972 | -53% | 1 | 1 | 0% | 1,988 | 1,591 | -20% | 0 | 0 | — |
case-19 | fail→fail | 11,512 | 10,978 | -5% | 1 | 1 | 0% | 2,207 | 2,757 | +25% | 0 | 0 | — |
case-20 | fail→fail | 9,129 | 6,942 | -24% | 1 | 1 | 0% | 1,687 | 1,961 | +16% | 0 | 0 | — |
case-21 | fail→fail | 5,552 | 5,180 | -7% | 1 | 1 | 0% | 1,313 | 1,588 | +21% | 0 | 0 | — |
case-22 | fail→fail | 13,901 | 12,681 | -9% | 1 | 1 | 0% | 2,666 | 3,672 | +38% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.