Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Delegate focused coding, research, review, or verification work to a bounded DevSpace subagent. Use when a task benefits from separate context, a specialist perspective, or a follow-up with the same worker.
.claude/skills/waishnav-subagents/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 1% | 0% |
Use the DevSpace CLI through the shell or process tool. Run commands from the project the subagent should work on.
Discover usable targets instead of guessing names:
bashdevspace agents targets --json
Configured profiles include a description and may define provider, model, effort, and task instructions. Choose a matching profile when one fits. Use a provider target when no profile fits or a specific provider is needed.
Usually rely on the target's configured model and effort. Pass --model or --effort only with a value supported by that provider. DevSpace passes these values through without translating them between providers.
Give the subagent a self-contained brief. Include the objective, relevant paths, constraints, decisions it needs from the current conversation, and the expected result. The subagent receives the brief and its profile instructions, not the parent conversation.
bashdevspace agents run <profile-or-provider> "<brief>" --json devspace agents run <profile-or-provider> --model <model> --effort <effort> "<brief>" --json
The result contains a DevSpace agent id and its current status. Execution continues independently, so retain the ID for later inspection or follow-up.
bashdevspace agents show <id> --json devspace agents continue <id> "<follow-up brief>" --json devspace agents ls --json
show waits briefly for active work, then returns the current status and anyavailable response or error.
continue gives the same subagent another turn with its existing providersession and context.
ls returns sessions belonging to the current project.Run devspace agents show <id> --json again later while the status is running. completed includes the response. failed includes a structured error, and stopped is terminal without a successful response. Continue an agent when its existing context is useful; start another agent for unrelated work.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 31,295 | 107,875 | +245% | 1 | 1 | 0% | 1,656 | 2,727 | +65% | 0 | 0 | — |
case-02 | fail→fail | 29,672 | 15,092 | -49% | 1 | 1 | 0% | 3,668 | 740 | -80% | 0 | 0 | — |
case-03 | fail→fail | 8,189 | 5,872 | -28% | 1 | 1 | 0% | 401 | 702 | +75% | 0 | 0 | — |
case-04 | fail→pass | 8,611 | 8,643 | +0% | 1 | 1 | 0% | 1,249 | 873 | -30% | 0 | 0 | — |
case-05 | fail→pass | 7,838 | 3,653 | -53% | 1 | 1 | 0% | 1,049 | 1,042 | -1% | 0 | 0 | — |
case-06 | fail→pass | 14,952 | 5,236 | -65% | 1 | 1 | 0% | 2,701 | 1,177 | -56% | 0 | 0 | — |
case-07 | fail→fail | 11,120 | 8,384 | -25% | 1 | 1 | 0% | 1,789 | 872 | -51% | 0 | 0 | — |
case-08 | pass→pass | 28,705 | 6,814 | -76% | 1 | 1 | 0% | 1,836 | 1,377 | -25% | 0 | 0 | — |
case-09 | fail→pass | 30,968 | 2,710 | -91% | 1 | 1 | 0% | 2,423 | 798 | -67% | 0 | 0 | — |
case-10 | fail→pass | 10,099 | 7,334 | -27% | 1 | 1 | 0% | 1,452 | 1,473 | +1% | 0 | 0 | — |
case-11 | fail→pass | 7,263 | 2,723 | -63% | 1 | 1 | 0% | 899 | 852 | -5% | 0 | 0 | — |
case-12 | pass→pass | 11,640 | 4,302 | -63% | 1 | 1 | 0% | 1,472 | 1,028 | -30% | 0 | 0 | — |
case-13 | pass→pass | 11,285 | 4,439 | -61% | 1 | 1 | 0% | 1,650 | 1,082 | -34% | 0 | 0 | — |
case-14 | fail→pass | 9,799 | 4,820 | -51% | 1 | 1 | 0% | 1,329 | 1,290 | -3% | 0 | 0 | — |
case-15 | fail→fail | 9,559 | 3,039 | -68% | 1 | 1 | 0% | 1,407 | 864 | -39% | 0 | 0 | — |
case-16 | pass→pass | 11,791 | 5,507 | -53% | 1 | 1 | 0% | 1,778 | 1,244 | -30% | 0 | 0 | — |
case-17 | pass→pass | 5,367 | 6,182 | +15% | 1 | 1 | 0% | 700 | 1,015 | +45% | 0 | 0 | — |
case-18 | fail→pass | 7,213 | 4,169 | -42% | 1 | 1 | 0% | 1,118 | 910 | -19% | 0 | 0 | — |
case-19 | fail→pass | 20,102 | 8,791 | -56% | 1 | 1 | 0% | 1,479 | 1,876 | +27% | 0 | 0 | — |
case-20 | fail→fail | 11,257 | 12,345 | +10% | 1 | 1 | 0% | 1,977 | 2,222 | +12% | 0 | 0 | — |
case-21 | pass→fail | 23,698 | 8,429 | -64% | 1 | 1 | 0% | 1,534 | 715 | -53% | 0 | 0 | — |
case-22 | fail→pass | 9,269 | 15,666 | +69% | 1 | 1 | 0% | 1,509 | 1,993 | +32% | 0 | 0 | — |
case-23 | fail→pass | 10,527 | 6,564 | -38% | 1 | 1 | 0% | 1,612 | 1,343 | -17% | 0 | 0 | — |
case-24 | fail→pass | 5,496 | 3,959 | -28% | 1 | 1 | 0% | 805 | 810 | +1% | 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. 24 cases were attempted, and 21 counted toward the lift figure. The other 3 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 +46 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.