Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when creating, validating, exporting, importing, or improving Runtype Full Product Object templates. Covers FPO and FPO template structure, template variables, pending secrets, setup-required auth, validate_product and validate_product_* checks, distributable product packaging, import readiness, and avoiding credential leakage.
.claude/skills/hashgraph-online-runtype-templates/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 79% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -14% | 0% |
Use this skill when a Runtype product should become a portable artifact that another workspace, team, customer, or marketplace user can import.
When MCP is available, fetch current schema docs before writing or validating templates:
get_platform_documentation(topic="product-schema")get_platform_documentation(topic="types-fpo")get_platform_documentation(topic="surface-types")get_platform_documentation(topic="flow-step-types")get_platform_documentation(topic="builtin-tools")get_platform_documentation(topic="orthogonal-tools")get_platform_documentation(topic="external-tools")runtype://types/fpo-templateruntype://types/surface-configsPrefer the live MCP docs over this skill's local fallback references. For the template wrapper itself, read runtype://types/fpo-template; for the wrapped productObject, read types-fpo because the current FPO default is version 2.0.
When a user provides a use.runtype.com/now?... quick-start URL, a runtype.com/examples/<slug> page, or a direct FPO/template JSON URL, do not open the login-gated dashboard page. Pass the original link directly to create_product_from_example as url (or use a known slug from list_example_templates).
variables argument. If requiredvariables are missing, ask for the returned values and retry.
allow_draft: true; use it only when the user intends toinstall that draft.
get_product_setup with the created product id. Complete automatable setup andreturn the remaining secret, OAuth, or surface-install URLs to the user.
execute_agent with amessages array of { "role": "user", "content": "..." } turns.
An FPO template wraps a product object and import-time variables:
productObject: the full product definition.template.variables: non-secret values the importer supplies.auth.setupRequired: true plus auth.secrets[]: pending-secret declarations.Template variables are for names, public URLs, recipient lists, selected models, and other non-secret configuration. Credentials must use the pending-secret pattern.
Use this pattern for every credential:
auth.secrets array.auth.setupRequired: true.{{secret:KEY}}.needs_configuration secret bindings.Do not use a template.variables entry with inputType: "secret" for credentials. That substitutes the value into the resolved product object.
validate_product on the resolved product object.validate_product_tool,validate_product_flow, validate_product_agent, and validate_product_surface.
runtype-admin instead.If the umbrella runtype skill is installed alongside this focused skill, its durable references provide deeper FPO template details. This skill must still work when installed by itself; prefer live MCP docs over local sibling files.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | fail→fail | 15,930 | 7,902 | -50% | 1 | 1 | 0% | 2,868 | 1,467 | -49% | 0 | 0 | — |
case-09 | fail→pass | 17,069 | 8,760 | -49% | 1 | 1 | 0% | 1,844 | 1,535 | -17% | 0 | 0 | — |
case-01 | fail→fail | 8,722 | 12,028 | +38% | 1 | 1 | 0% | 1,332 | 1,442 | +8% | 0 | 0 | — |
case-02 | fail→pass | 20,258 | 21,185 | +5% | 1 | 1 | 0% | 3,073 | 5,489 | +79% | 0 | 0 | — |
case-03 | fail→fail | 5,908 | 17,986 | +204% | 1 | 1 | 0% | 497 | 3,161 | +536% | 0 | 0 | — |
case-04 | fail→pass | 9,169 | 3,526 | -62% | 1 | 1 | 0% | 1,450 | 1,577 | +9% | 0 | 0 | — |
case-05 | fail→fail | 21,081 | 11,032 | -48% | 1 | 1 | 0% | 2,753 | 1,871 | -32% | 0 | 0 | — |
case-06 | fail→pass | 20,283 | 7,462 | -63% | 1 | 1 | 0% | 2,099 | 2,378 | +13% | 0 | 0 | — |
case-07 | pass→pass | 19,413 | 10,527 | -46% | 1 | 1 | 0% | 2,325 | 1,954 | -16% | 0 | 0 | — |
case-08 | fail→pass | 13,401 | 8,918 | -33% | 1 | 1 | 0% | 1,963 | 1,695 | -14% | 0 | 0 | — |
case-10 | pass→pass | 13,919 | 6,419 | -54% | 1 | 1 | 0% | 1,465 | 2,277 | +55% | 0 | 0 | — |
case-11 | fail→pass | 16,167 | 4,836 | -70% | 1 | 1 | 0% | 1,762 | 1,602 | -9% | 0 | 0 | — |
case-12 | fail→pass | 26,958 | 8,151 | -70% | 1 | 1 | 0% | 2,381 | 1,464 | -39% | 0 | 0 | — |
case-13 | fail→pass | 16,571 | 7,674 | -54% | 1 | 1 | 0% | 1,786 | 2,117 | +19% | 0 | 0 | — |
case-14 | fail→pass | 15,871 | 8,448 | -47% | 1 | 1 | 0% | 1,682 | 2,157 | +28% | 0 | 0 | — |
case-16 | pass→pass | 16,176 | 6,354 | -61% | 1 | 1 | 0% | 2,109 | 2,055 | -3% | 0 | 0 | — |
case-17 | fail→pass | 26,315 | 2,674 | -90% | 1 | 1 | 0% | 4,058 | 1,410 | -65% | 0 | 0 | — |
case-18 | fail→fail | 17,601 | 3,055 | -83% | 1 | 1 | 0% | 1,843 | 1,556 | -16% | 0 | 0 | — |
case-19 | fail→fail | 17,033 | 7,216 | -58% | 1 | 1 | 0% | 1,725 | 1,213 | -30% | 0 | 0 | — |
case-20 | pass→pass | 11,732 | 19,677 | +68% | 1 | 1 | 0% | 960 | 2,818 | +194% | 0 | 0 | — |
case-21 | pass→pass | 20,251 | 13,117 | -35% | 1 | 1 | 0% | 2,395 | 2,435 | +2% | 0 | 0 | — |
case-22 | pass→pass | 17,458 | 31,799 | +82% | 1 | 1 | 0% | 3,368 | 3,390 | +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. 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 +45 percentage points is the difference between those two pass rates over the 21 comparable cases.
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.