Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Delegate tasks to OpenAI Codex CLI and Google Antigravity CLI from Claude Code. Say 'check with codex' or 'ask gemini for a second opinion' and Claude runs the other agent, keeps a topic-aware session, and critiques the result.
.claude/skills/davepoon-dispatch/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 11% | 0% |
Delegate tasks to external AI CLIs from inside Claude Code. Say "check with codex", "ask gemini for a second opinion", or "validate this before I merge" and Claude runs the other agent, keeps a topic-aware conversation, and critiques the result rather than echoing it.
Supports OpenAI Codex CLI and Google Antigravity CLI (multi-model: Gemini, Claude, GPT-OSS). No API keys — both CLIs use their own auth (OAuth / Google sign-in).
Source: sparklingneuronics/sparkling-skills (MIT)
User: ask codex to review my uncommitted changes before I merge
Claude: [runs codex review --uncommitted, synthesizes findings]
Codex flagged three things. I agree with #1 (missing null check)
but #2 and #3 are false positives. Want me to fix the null check?User: ask gemini to challenge this architecture
Claude: [runs agy with Gemini 3.1 Pro, compares with own analysis]
Gemini raises a valid concern about hot-partition risk that I
hadn't flagged. Its other two points don't apply here.User: continue with codex — is the top issue real?
Claude: [resumes the same codex session with a delta bridge]
Codex confirms the null-check issue — it traced the code path.codex CLI + codex login (OAuth)agy CLI + Google sign-in (free account)Install: claude install-plugin sparklingneuronics/sparkling-skills
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | fail→fail | 8,632 | 2,368 | -73% | 1 | 1 | 0% | 1,269 | 876 | -31% | 0 | 0 | — |
case-01 | fail→fail | 8,289 | 5,973 | -28% | 1 | 1 | 0% | 963 | 1,265 | +31% | 0 | 0 | — |
case-02 | fail→fail | 6,910 | 3,902 | -44% | 1 | 1 | 0% | 942 | 871 | -8% | 0 | 0 | — |
case-03 | fail→fail | 3,685 | 9,688 | +163% | 1 | 1 | 0% | 464 | 1,121 | +142% | 0 | 0 | — |
case-04 | fail→fail | 6,736 | 12,476 | +85% | 1 | 1 | 0% | 896 | 927 | +3% | 0 | 0 | — |
case-05 | fail→fail | 11,523 | 5,653 | -51% | 1 | 1 | 0% | 1,945 | 1,392 | -28% | 0 | 0 | — |
case-06 | fail→fail | 5,698 | 6,328 | +11% | 1 | 1 | 0% | 932 | 757 | -19% | 0 | 0 | — |
case-07 | fail→pass | 8,479 | 3,432 | -60% | 1 | 1 | 0% | 1,389 | 1,189 | -14% | 0 | 0 | — |
case-08 | fail→pass | 7,161 | 2,611 | -64% | 1 | 1 | 0% | 1,217 | 1,039 | -15% | 0 | 0 | — |
case-09 | fail→fail | 15,765 | 9,425 | -40% | 1 | 1 | 0% | 1,632 | 1,437 | -12% | 0 | 0 | — |
case-10 | fail→pass | 7,849 | 7,517 | -4% | 1 | 1 | 0% | 1,118 | 1,132 | +1% | 0 | 0 | — |
case-11 | fail→fail | 6,797 | 7,180 | +6% | 1 | 1 | 0% | 987 | 1,542 | +56% | 0 | 0 | — |
case-13 | pass→fail | 9,194 | 7,000 | -24% | 1 | 1 | 0% | 1,482 | 1,029 | -31% | 0 | 0 | — |
case-14 | fail→fail | 15,462 | 6,916 | -55% | 1 | 1 | 0% | 2,223 | 993 | -55% | 0 | 0 | — |
case-15 | fail→pass | 15,822 | 12,662 | -20% | 1 | 1 | 0% | 2,858 | 2,560 | -10% | 0 | 0 | — |
case-16 | fail→fail | 9,352 | 4,963 | -47% | 1 | 1 | 0% | 1,316 | 788 | -40% | 0 | 0 | — |
case-17 | fail→fail | 10,774 | 7,710 | -28% | 1 | 1 | 0% | 1,608 | 924 | -43% | 0 | 0 | — |
case-18 | fail→pass | 9,966 | 7,732 | -22% | 1 | 1 | 0% | 1,562 | 1,740 | +11% | 0 | 0 | — |
case-19 | pass→pass | 3,450 | 4,885 | +42% | 1 | 1 | 0% | 447 | 1,348 | +202% | 0 | 0 | — |
case-20 | pass→pass | 3,637 | 3,481 | -4% | 1 | 1 | 0% | 536 | 1,136 | +112% | 0 | 0 | — |
case-21 | pass→pass | 7,472 | 5,678 | -24% | 1 | 1 | 0% | 1,503 | 1,706 | +14% | 0 | 0 | — |
case-22 | fail→pass | 2,731 | 2,849 | +4% | 1 | 1 | 0% | 293 | 1,024 | +249% | 0 | 0 | — |
case-23 | pass→pass | 10,363 | 8,248 | -20% | 1 | 1 | 0% | 1,890 | 2,007 | +6% | 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. 23 cases were attempted, and 19 counted toward the lift figure. The other 4 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +22 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
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.