Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Validate a Claude Code plugin structure, frontmatter, and MCP tool references
.claude/skills/ruvnet-validate-plugin/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -31% | 0% |
Validate that a plugin follows the correct Claude Code plugin format.
After creating or modifying a plugin, run validation to catch structural issues before publishing.
.claude-plugin/plugin.json exists at plugin rootskills/<name>/SKILL.md is a valid skill (Claude Code auto-discovers from directory; plugin.json MUST NOT list a skills array)commands/<name>.md is a valid command (auto-discovered; no commands array in plugin.json)agents/<name>.md is a valid agent (auto-discovered; no agents array in plugin.json)skills, commands, or agents arrays in plugin.json is a validation error (they cause Claude Code to reject the plugin)name, description, and allowed-tools (no wildcards)name, description, and model.claude-plugin/allowed-tools are valid mcp__plugin_ruflo-core_ruflo__* identifiersplugin.json and assert no skills / commands / agents arrays presentskills/*/SKILL.md, commands/*.md, agents/*.md and validate each frontmatterallowed-tools has no wildcards| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,379 | 2,118 | -67% | 1 | 1 | 0% | 1,148 | 698 | -39% | 0 | 0 | — |
case-02 | pass→fail | 16,548 | 2,112 | -87% | 1 | 1 | 0% | 3,377 | 680 | -80% | 0 | 0 | — |
case-03 | fail→fail | 3,276 | 6,152 | +88% | 1 | 1 | 0% | 183 | 987 | +439% | 0 | 0 | — |
case-04 | fail→pass | 10,278 | 3,862 | -62% | 1 | 1 | 0% | 1,889 | 1,223 | -35% | 0 | 0 | — |
case-05 | fail→pass | 11,637 | 3,214 | -72% | 1 | 1 | 0% | 1,882 | 1,132 | -40% | 0 | 0 | — |
case-06 | fail→pass | 11,487 | 3,242 | -72% | 1 | 1 | 0% | 1,988 | 1,110 | -44% | 0 | 0 | — |
case-07 | fail→pass | 10,631 | 3,850 | -64% | 1 | 1 | 0% | 1,819 | 1,207 | -34% | 0 | 0 | — |
case-08 | fail→pass | 10,249 | 4,014 | -61% | 1 | 1 | 0% | 1,930 | 1,332 | -31% | 0 | 0 | — |
case-09 | fail→pass | 9,608 | 3,781 | -61% | 1 | 1 | 0% | 1,713 | 1,214 | -29% | 0 | 0 | — |
case-10 | fail→pass | 14,634 | 2,427 | -83% | 1 | 1 | 0% | 2,382 | 873 | -63% | 0 | 0 | — |
case-11 | pass→pass | 9,424 | 3,346 | -64% | 1 | 1 | 0% | 1,600 | 1,083 | -32% | 0 | 0 | — |
case-12 | fail→pass | 5,927 | 2,571 | -57% | 1 | 1 | 0% | 1,186 | 961 | -19% | 0 | 0 | — |
case-13 | pass→pass | 10,289 | 1,928 | -81% | 1 | 1 | 0% | 2,122 | 864 | -59% | 0 | 0 | — |
case-14 | fail→pass | 8,268 | 3,074 | -63% | 1 | 1 | 0% | 1,579 | 1,129 | -28% | 0 | 0 | — |
case-15 | fail→pass | 8,477 | 2,467 | -71% | 1 | 1 | 0% | 1,347 | 926 | -31% | 0 | 0 | — |
case-16 | pass→pass | 9,900 | 2,519 | -75% | 1 | 1 | 0% | 2,008 | 940 | -53% | 0 | 0 | — |
case-17 | fail→pass | 7,169 | 3,408 | -52% | 1 | 1 | 0% | 1,268 | 1,050 | -17% | 0 | 0 | — |
case-18 | fail→pass | 6,907 | 2,917 | -58% | 1 | 1 | 0% | 1,192 | 1,017 | -15% | 0 | 0 | — |
case-19 | fail→pass | 10,261 | 3,429 | -67% | 1 | 1 | 0% | 1,813 | 1,231 | -32% | 0 | 0 | — |
case-20 | pass→pass | 7,573 | 5,381 | -29% | 1 | 1 | 0% | 1,554 | 1,624 | +5% | 0 | 0 | — |
case-21 | pass→pass | 2,551 | 3,349 | +31% | 1 | 1 | 0% | 480 | 1,170 | +144% | 0 | 0 | — |
case-22 | pass→pass | 2,946 | 2,355 | -20% | 1 | 1 | 0% | 518 | 964 | +86% | 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 +55 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.