Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Submit, browse, install, edit, archive, restore, transfer, and version MCPs, skills, hooks, prompts, and sandboxes in the Observal registry, and get components recommended for the current user. Use when the user wants to submit a component, install one, edit a draft, publish a new version, browse the component library, or asks what they should install or are missing.
.claude/skills/observal-observal-registry/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 5% | 0% |
--output json whenever supported. Parse list results from items and pagination fields.--help when any path, flag, enum, or payload shape is uncertain.qualified_name values. Never automate with row numbers or ambiguous bare names.deployment.public_registry_enabled is enabled; it is disabled by default on self-hosted deployments. Public list, show, install, and prompt render commands use https://public.observal.io by default. Authenticate before submitting, editing, reviewing, rating, or accessing private team content.| User intent | Read | | --- | --- | | Find, inspect, recommend, or install components | Discovery and installation | | Submit one component or a mixed bulk file | Component submission | | Edit, version, archive, restore, transfer, or manage co-authors | Registry lifecycle |
Read only the selected reference, and read it completely before executing.
personalized: false means popularity fallback, not a personal recommendation.TEAM_HANDLE/ITEM_SLUG for direct references.Report component type, canonical identity, version, status, target harness or scope when applicable, warnings, and any review or setup step still required.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 16,576 | 18,925 | +14% | 1 | 1 | 0% | 2,058 | 1,042 | -49% | 0 | 0 | — |
case-02 | fail→fail | 11,583 | 18,066 | +56% | 1 | 1 | 0% | 933 | 982 | +5% | 0 | 0 | — |
case-03 | fail→fail | 13,590 | 14,721 | +8% | 1 | 1 | 0% | 1,209 | 907 | -25% | 0 | 0 | — |
case-04 | fail→fail | 12,339 | 18,773 | +52% | 1 | 1 | 0% | 1,553 | 1,007 | -35% | 0 | 0 | — |
case-05 | fail→pass | 13,420 | 7,221 | -46% | 1 | 1 | 0% | 1,187 | 820 | -31% | 0 | 0 | — |
case-06 | pass→pass | 18,779 | 10,331 | -45% | 1 | 1 | 0% | 1,971 | 1,225 | -38% | 0 | 0 | — |
case-07 | fail→pass | 9,829 | 10,861 | +10% | 1 | 1 | 0% | 1,469 | 1,367 | -7% | 0 | 0 | — |
case-08 | fail→pass | 14,654 | 3,369 | -77% | 1 | 1 | 0% | 1,461 | 1,028 | -30% | 0 | 0 | — |
case-09 | pass→pass | 14,633 | 3,843 | -74% | 1 | 1 | 0% | 1,252 | 999 | -20% | 0 | 0 | — |
case-10 | fail→pass | 20,987 | 8,805 | -58% | 1 | 1 | 0% | 2,293 | 1,072 | -53% | 0 | 0 | — |
case-11 | pass→pass | 13,722 | 3,127 | -77% | 1 | 1 | 0% | 1,268 | 1,045 | -18% | 0 | 0 | — |
case-12 | pass→pass | 15,784 | 11,709 | -26% | 1 | 1 | 0% | 1,672 | 1,508 | -10% | 0 | 0 | — |
case-13 | pass→pass | 12,430 | 8,367 | -33% | 1 | 1 | 0% | 1,020 | 976 | -4% | 0 | 0 | — |
case-14 | fail→pass | 52,885 | 15,469 | -71% | 1 | 1 | 0% | 2,139 | 2,255 | +5% | 0 | 0 | — |
case-15 | pass→pass | 30,538 | 5,163 | -83% | 1 | 1 | 0% | 1,931 | 1,214 | -37% | 0 | 0 | — |
case-16 | pass→fail | 13,631 | 2,886 | -79% | 1 | 1 | 0% | 2,011 | 979 | -51% | 0 | 0 | — |
case-17 | fail→pass | 31,366 | 4,383 | -86% | 1 | 1 | 0% | 4,947 | 1,069 | -78% | 0 | 0 | — |
case-18 | pass→pass | 15,287 | 2,813 | -82% | 1 | 1 | 0% | 2,056 | 997 | -52% | 0 | 0 | — |
case-19 | pass→pass | 15,784 | 2,672 | -83% | 1 | 1 | 0% | 2,107 | 894 | -58% | 0 | 0 | — |
case-20 | pass→pass | 10,437 | 5,051 | -52% | 1 | 1 | 0% | 1,622 | 1,094 | -33% | 0 | 0 | — |
case-21 | pass→pass | 19,236 | 5,833 | -70% | 1 | 1 | 0% | 1,132 | 1,398 | +23% | 0 | 0 | — |
case-22 | pass→pass | 13,721 | 54,711 | +299% | 1 | 1 | 0% | 2,430 | 2,888 | +19% | 0 | 0 | — |
case-23 | pass→pass | 9,248 | 8,749 | -5% | 1 | 1 | 0% | 1,086 | 1,621 | +49% | 0 | 0 | — |
case-24 | pass→pass | 6,811 | 8,633 | +27% | 1 | 1 | 0% | 1,255 | 1,998 | +59% | 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 +21 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/17/2026 | +23% |
| gemini-3.6-flash | verified | 8/14/2026 | +24% |
Other measured skills in the registry, with their headline benchmark lift.