Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Diagnose a named CCB agent by combining authoritative runtime and job lineage with deep read-only pane inspection, apply bounded recovery when evidence supports it, verify the result, and request authorization before submitting a redacted GitHub issue. Use for `$ccb_diagnose agentname`, `$ccb-diagnose agentname`, or reports that a CCB agent is stuck, disconnected, not continuing, not replying, or showing provider errors.
.claude/skills/seemseam-ccb-diagnose/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -45% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -87% | 0% |
Diagnose exactly one current mounted agent. Start with CCB authority:
bashcommand ccb ping "$AGENT" command ccb ps command ccb queue --detail "$AGENT" command ccb pend --inbox --detail "$AGENT"
Use ccb trace for a current lineage id and ccb doctor logs for provider/API evidence. For DSH, the pane is only the managed host process/log surface: pane text is not prompt, reply, or completion authority. Native authority is the exact DSH session/RPC event history and turn/end reason recorded by CCB.
Do not read credentials, mutate tmux directly, restart all agents, or submit a GitHub issue without showing a redacted proposal and receiving explicit authorization. Apply only a bounded supported repair, then re-check the same runtime and lineage evidence.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,056 | 14,499 | +80% | 1 | 1 | 0% | 228 | 546 | +139% | 0 | 0 | — |
case-02 | fail→fail | 11,951 | 4,591 | -62% | 1 | 1 | 0% | 1,727 | 596 | -65% | 0 | 0 | — |
case-03 | fail→fail | 8,151 | 6,756 | -17% | 1 | 1 | 0% | 496 | 595 | +20% | 0 | 0 | — |
case-04 | fail→fail | 11,983 | 9,044 | -25% | 1 | 1 | 0% | 1,783 | 775 | -57% | 0 | 0 | — |
case-05 | fail→fail | 8,423 | 6,578 | -22% | 1 | 1 | 0% | 1,107 | 623 | -44% | 0 | 0 | — |
case-06 | pass→fail | 9,607 | 7,497 | -22% | 1 | 1 | 0% | 1,445 | 749 | -48% | 0 | 0 | — |
case-07 | fail→fail | 14,180 | 9,444 | -33% | 1 | 1 | 0% | 1,837 | 821 | -55% | 0 | 0 | — |
case-08 | fail→fail | 7,820 | 16,600 | +112% | 1 | 1 | 0% | 1,188 | 841 | -29% | 0 | 0 | — |
case-09 | fail→fail | 5,056 | 8,098 | +60% | 1 | 1 | 0% | 777 | 653 | -16% | 0 | 0 | — |
case-10 | fail→fail | 10,273 | 8,648 | -16% | 1 | 1 | 0% | 1,355 | 602 | -56% | 0 | 0 | — |
case-11 | fail→pass | 7,313 | 3,966 | -46% | 1 | 1 | 0% | 1,001 | 807 | -19% | 0 | 0 | — |
case-12 | fail→fail | 14,561 | 2,938 | -80% | 1 | 1 | 0% | 2,438 | 571 | -77% | 0 | 0 | — |
case-13 | fail→pass | 15,446 | 11,017 | -29% | 1 | 1 | 0% | 2,712 | 1,492 | -45% | 0 | 0 | — |
case-14 | fail→pass | 5,764 | 1,767 | -69% | 1 | 1 | 0% | 1,009 | 514 | -49% | 0 | 0 | — |
case-15 | fail→pass | 8,822 | 1,931 | -78% | 1 | 1 | 0% | 1,513 | 490 | -68% | 0 | 0 | — |
case-16 | fail→fail | 11,154 | 6,534 | -41% | 1 | 1 | 0% | 1,645 | 580 | -65% | 0 | 0 | — |
case-17 | pass→fail | 10,777 | 13,363 | +24% | 1 | 1 | 0% | 1,661 | 544 | -67% | 0 | 0 | — |
case-18 | fail→fail | 8,496 | 8,779 | +3% | 1 | 1 | 0% | 1,124 | 710 | -37% | 0 | 0 | — |
case-19 | fail→pass | 22,913 | 2,403 | -90% | 1 | 1 | 0% | 4,120 | 518 | -87% | 0 | 0 | — |
case-20 | fail→pass | 13,076 | 4,959 | -62% | 1 | 1 | 0% | 1,883 | 703 | -63% | 0 | 0 | — |
case-25 | pass→fail | 9,338 | 7,346 | -21% | 1 | 1 | 0% | 1,809 | 499 | -72% | 0 | 0 | — |
case-21 | pass→fail | 14,664 | 11,866 | -19% | 1 | 1 | 0% | 2,188 | 1,712 | -22% | 0 | 0 | — |
case-22 | fail→pass | 11,913 | 3,331 | -72% | 1 | 1 | 0% | 1,548 | 566 | -63% | 0 | 0 | — |
case-23 | pass→fail | 16,323 | 6,902 | -58% | 1 | 1 | 0% | 2,415 | 635 | -74% | 0 | 0 | — |
case-24 | pass→fail | 13,941 | 7,646 | -45% | 1 | 1 | 0% | 2,047 | 524 | -74% | 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. 25 cases were attempted, and 10 counted toward the lift figure. The other 15 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 +4 percentage points is the difference between those two pass rates over the 10 comparable cases. 8 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.