Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Org-level agent team designer via `epic team` CLI (cross-project, append-merge). Subcommands: list, show, sync, link, unlink, history. Use when setting up agents, designing teams, or /spec yields 3+ requirements.
.claude/skills/hashgraph-online-team/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -70% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -51% | 0% |
This skill is a thin wrapper around the epic team CLI.
Run in terminal:
epic teamepic team handles the full interactive flow:
HARNESS_ORG env → prompt → default "epic")~/.harness/orgs/.claude/agents/{team}/ with ## Team Context injectedFor the full spec see docs/research/team-spec.md.
epic team list # list teams in current org
epic team show {team} # config + agents + mission
epic team show {team} --playbook # full accumulated playbook
epic team sync {team} # re-copy agents to .claude/agents/
epic team link {team} # attach existing team (skip design)
epic team unlink {team} # remove .claude/agents/{team}/
epic team history {team} {agent} # show .history/ entries| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 9,464 | 19,185 | +103% | 1 | 1 | 0% | 1,321 | 834 | -37% | 0 | 0 | — |
case-02 | fail→pass | 16,867 | 3,296 | -80% | 1 | 1 | 0% | 1,768 | 606 | -66% | 0 | 0 | — |
case-03 | fail→pass | 15,175 | 8,979 | -41% | 1 | 1 | 0% | 1,577 | 707 | -55% | 0 | 0 | — |
case-04 | pass→fail | 13,866 | 13,952 | +1% | 1 | 1 | 0% | 2,326 | 2,255 | -3% | 0 | 0 | — |
case-05 | pass→pass | 6,167 | 8,852 | +44% | 1 | 1 | 0% | 877 | 727 | -17% | 0 | 0 | — |
case-06 | pass→pass | 7,619 | 11,643 | +53% | 1 | 1 | 0% | 1,174 | 1,198 | +2% | 0 | 0 | — |
case-07 | pass→pass | 14,594 | 7,929 | -46% | 1 | 1 | 0% | 1,276 | 671 | -47% | 0 | 0 | — |
case-08 | fail→pass | 18,334 | 7,749 | -58% | 1 | 1 | 0% | 2,059 | 623 | -70% | 0 | 0 | — |
case-09 | fail→pass | 12,241 | 3,224 | -74% | 1 | 1 | 0% | 1,722 | 847 | -51% | 0 | 0 | — |
case-10 | fail→pass | 8,542 | 7,646 | -10% | 1 | 1 | 0% | 1,254 | 616 | -51% | 0 | 0 | — |
case-11 | fail→pass | 17,111 | 1,958 | -89% | 1 | 1 | 0% | 1,657 | 622 | -62% | 0 | 0 | — |
case-12 | fail→pass | 19,122 | 2,171 | -89% | 1 | 1 | 0% | 2,346 | 579 | -75% | 0 | 0 | — |
case-13 | fail→pass | 14,147 | 7,139 | -50% | 1 | 1 | 0% | 1,626 | 511 | -69% | 0 | 0 | — |
case-14 | fail→pass | 34,847 | 7,169 | -79% | 1 | 1 | 0% | 5,180 | 534 | -90% | 0 | 0 | — |
case-15 | fail→pass | 13,427 | 1,907 | -86% | 1 | 1 | 0% | 1,435 | 520 | -64% | 0 | 0 | — |
case-16 | fail→pass | 20,594 | 1,453 | -93% | 1 | 1 | 0% | 2,578 | 466 | -82% | 0 | 0 | — |
case-17 | fail→pass | 16,850 | 1,576 | -91% | 1 | 1 | 0% | 1,602 | 457 | -71% | 0 | 0 | — |
case-18 | fail→pass | 12,121 | 7,612 | -37% | 1 | 1 | 0% | 1,712 | 601 | -65% | 0 | 0 | — |
case-19 | pass→pass | 20,562 | 8,372 | -59% | 1 | 1 | 0% | 1,997 | 636 | -68% | 0 | 0 | — |
case-20 | pass→pass | 11,039 | 3,261 | -70% | 1 | 1 | 0% | 1,634 | 845 | -48% | 0 | 0 | — |
case-21 | pass→pass | 14,892 | 8,380 | -44% | 1 | 1 | 0% | 1,502 | 665 | -56% | 0 | 0 | — |
case-22 | fail→pass | 4,897 | 6,982 | +43% | 1 | 1 | 0% | 736 | 508 | -31% | 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, and 21 counted toward the lift figure. The other 1 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 +59 percentage points is the difference between those two pass rates over the 21 comparable cases. 1 case got worse with the skill loaded, and it is 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.