AGENT SKILL

runtype-templates

Create, validate, or share distributable Runtype Full Product Object (FPO) templates with import-time variables and pending secrets, or install a product from a runtype.com example, a use.runtype.com/now link, or a template URL.

License 许可
Apache-2.0
最近更新
2026-11-13
Stars
★ 1.6k
Forks
483

01 前言 Front Matter

name
runtype-templates
description
Create, validate, or share distributable Runtype Full Product Object (FPO) templates with import-time variables and pending secrets, or install a product from a runtype.com example, a use.runtype.com/now link, or a template URL.
argument-hint
[FPO or template task]
user-invocable
true

02 资源文件 1

03 SKILL.md 正文

Runtype Templates

Use this skill when a Runtype product should become a portable Full Product Object (FPO) template that another workspace, team, customer, or marketplace user can import. It also covers installing a product from a shared example or template link.

Source of truth

When the Runtype MCP server is available, read the schema docs before you write or validate a template.

Always read:

  • get_platform_documentation(topic="types-fpo-template") for the template wrapper.
  • get_platform_documentation(topic="types-fpo") for the wrapped productObject. Use FPO version 2.0 for new templates.
  • get_platform_documentation(topic="product-schema") for the JSON Schema of the product object.

Read these when the template contains that part:

  • Surfaces: surface-types and types-surface-configs.
  • Flows: flow-step-types.
  • Tools: builtin-tools, orthogonal-tools, and external-tools.

For prose guidance, call search_documentation for "FPO templates" or read the public guides: FPO templates and Importing products.

Template shape

An FPO template has three top-level keys:

  • version: the template format version, "1.0" or "1.1". This is separate from productObject.version, which is "2.0" for new templates.
  • productObject: the full product definition.
  • template.variables: the non-secret values the importer supplies.

A template must declare at least one variable or at least one pending secret: a tool with auth.setupRequired: true and a non-empty auth.secrets array, a top-level productObject.secrets[] entry, or a {{secret:KEY}} reference anywhere in the product.

Use variables for names, public URLs, recipient lists, selected models, and other non-secret configuration. Every credential uses the pending-secret pattern.

Variables

Each variable needs key, label, inputType, and required:

{
  "key": "companyName",
  "label": "Company name",
  "inputType": "text",
  "required": true,
  "defaultValue": "Acme"
}
  • inputType is one of text, textarea, url, secret, select, number, or boolean.
  • A select variable needs a non-empty options array of { "label", "value" }. Only select variables can have options.
  • A template can declare up to 50 variables.
  • Set defaults with defaultValue. Inline {{key|default}} syntax is rejected.
  • When a string field is exactly {{key}} and the variable is number or boolean, the field keeps that type. Every other substitution produces a string.
  • A typed field (a URL such as externalConfig.endpoint, a number such as config.loopConfig.maxTurns, an enum, or a datetime such as a schedule's runAt) can hold a variable. Validation checks such fields after substitution, so a bad default or import value fails there instead.

Runtime tokens and template variables

Import treats every bare {{name}} in productObject as a template variable. That includes runtime tokens such as a step output in a prompt. Only these references pass through to run time without a declaration:

  • {{secret:KEY}}
  • {{flow:name}}, which resolves to the flow variable name at run time
  • System and message variables, such as {{_record.metadata.field}}, {{_now}}, {{userMessage}}, and {{messages}}
  • Host-resolved variables: {{_identity.token}} in a tool or MCP header, and {{_agent.*}} / {{_organization.id}} in a memory profileTemplate, including a fallback such as {{_agent.alias || "live"}}

When you turn a working product into a template, rewrite runtime flow tokens with the flow: prefix. For example, change Summarize {{research}} to Summarize {{flow:research}}. Do not declare a runtime token as a template variable, because import replaces it with a fixed value.

If a field cannot take the flow: prefix, such as a Liquid template body, declare a template variable whose defaultValue is the runtime text, and reference that variable in the field.

Optional bundled resources

A template can also ship these productObject arrays. Import creates each resource:

  • schedules[]: cron or one-time runs that target a capability through capabilityId. A scheduled capability does not need a type: "schedule" surface.
  • evals[]: eval suites with their cases.
  • skills[]: agent skills, bound to agent capabilities through bindTo.

For the field tables, see the FPO templates guide.

Pending secrets

Use this pattern for every credential:

  1. Declare the key on the target tool's auth.secrets array.
  2. Set auth.setupRequired: true, plus auth.type and auth.setupInstructions.
  3. Reference the key in the tool config as {{secret:KEY}}.
  4. Leave the value out of the template. Import creates the secret as unconfigured, and the importer fills it in after import.
{
  "config": {
    "headers": { "Authorization": "Bearer {{secret:ACME_API_KEY}}" }
  },
  "auth": {
    "type": "api_key",
    "setupRequired": true,
    "secrets": [{ "key": "ACME_API_KEY", "required": true }],
    "setupInstructions": {
      "summary": "Add your Acme API key.",
      "steps": ["Create an API key in Acme.", "Add it as a secret after import."]
    }
  }
}

Each auth.secrets entry is an object with key and required, not a bare string.

When the credential belongs to the product rather than one tool (an MCP server header, an external agent's auth, a model key), declare it in top-level productObject.secrets[] ({ "id", "key", "description"?, "required"? }) and reference it as {{secret:KEY}} wherever it is used. Import lists it with the other pending secrets.

Do not use a variable with inputType: "secret" for credentials. Import writes that value into the product object as plain text. A secret variable cannot have a defaultValue.

Validation workflow

  1. Validate the template document with runtype validate-product ./template.json --template or POST https://api.runtype.com/v1/public/products/validate-template. Fix every error, such as UNDECLARED_TEMPLATE_VARIABLE, INLINE_TEMPLATE_DEFAULT_NOT_ALLOWED, and DUPLICATE_TEMPLATE_VARIABLE. Review warnings such as UNUSED_TEMPLATE_VARIABLE and TEMPLATE_VARIABLE_SECRET_IS_UNSAFE.
  2. Resolve template defaults, or supply test values for required variables.
  3. Run validate_product on the resolved product object.
  4. Run the focused validators for changed parts: validate_product_tool, validate_product_flow, validate_product_agent, and validate_product_surface.
  5. Confirm that every route references a real capability id.
  6. Confirm that every secret reference has a matching pending-secret declaration.
  7. Confirm that importer-visible variables are few and meaningful.

Share a template

  1. Host the template JSON at a public HTTPS URL, such as a raw GitHub file URL.

  2. URL-encode that URL and share https://use.runtype.com/now?from=ENCODED_TEMPLATE_URL.

  3. Optional: Prefill a variable with &templateParam%3AVARIABLE_KEY=VALUE. The import page ignores prefilled values for secret variables.

  4. Optional: Add a button to a repository README:

    [Deploy to Runtype](https://use.runtype.com/now?from=https%3A%2F%2Fexample.com%2FYOUR_TEMPLATE.json)
    
  5. Test the link: pass it to create_product_from_example as url and confirm the product is created.

Set up a product from a link

When a user provides a use.runtype.com/now?... link, a runtype.com/examples/<slug> page, or a direct FPO or template JSON URL, do not open the dashboard page, which requires sign-in. Pass the original link to create_product_from_example as url, or pass a known slug from list_example_templates.

  1. Supply import-time values in the tool's variables argument. Omitted keys fall back to templateParam values in the link, then to template defaults. If a required variable still has no value, the tool returns the variable specs and creates nothing. Ask the user for those values, then call the tool again with them in variables.
  2. Draft examples require allow_draft: true. Use it only when the user intends to install that draft.
  3. Call get_product_setup with the created product id. Complete the steps you can automate, and give the user the remaining secret, OAuth, or surface-install links. Never ask the user to paste secret values in chat. Give them the dashboard link from get_product_setup or get_secret_intake_manifest, then confirm with check_secrets.
  4. Test the product at its user-facing layer. For an agent, call execute_agent with a messages array of { "role": "user", "content": "..." } turns.

The tool rejects some links:

  • A link with ?import= or ?sessionId= points to a temporary import preview. Ask for the original source: an example slug, a ?session= link, or the document URL.
  • A link with ?template= points to a built-in dashboard template. Open it in the dashboard, or build the product with create_product.
  • A /now link with none of example, session, templateUrl, or from has nothing to import. Ask the user for the full link.

Anti-patterns

  • Hardcoded API keys, tokens, client secrets, or bearer values.
  • Too many variables. Every variable adds setup work for the importer.
  • Workspace ids, runtime ids, or surface keys as template variables.
  • Bare runtime tokens such as {{research}} left in the product object.
  • Partial product schemas with a note that validation still needs work.
  • Templates for one-off private account updates. Use the runtype-admin skill instead.

For platform scoping and account setup, use the runtype skill.

04 同仓库更多 Skills

hashgraph-online
hashgraph-online/awesome-codex-plugins

migrate-to-e2e

Moves an existing Playwright, Cypress, Selenium, Appium, Detox, Maestro, or TesterArmy suite to e2e one slice at a time, with the old suite running beside the new one until each ported test is proven. Keeps a MIGRATION.md progress file so any later session resumes where the last stopped. Optionally plants a defect to show both suites catch it. Use for /migrate-to-e2e, "migrate our tests to e2e", "port these Maestro flows", "move our TesterArmy tests", or when a project has an e2e config and an older suite still running.

★ 1.6k
hashgraph-online
hashgraph-online/awesome-codex-plugins

e2e

Writes, runs, and debugs end-to-end tests for web and mobile apps with e2e, the agentic test runner, which mixes agent.act goals with exact locator assertions on a Playwright browser or an iOS or Android device. Also drives a project's running app live over the e2e MCP server to look at a screen, check a UI change, take a screenshot or record a video, and try a locator; and runs bug bashes, parallel e2e explore runs with each bug proven by a repro test. Use when a project depends on e2e or has an e2e.config.ts, when asked for end-to-end, browser, mobile, or agentic UI tests, to check a change in the running app, to QA or bug bash an app or branch, or when an e2e run fails.

★ 1.6k
hashgraph-online
hashgraph-online/awesome-codex-plugins

create-verification-skill

Generate a project-local verify-<app> skill on top of e2e, so any coding agent can launch the app, check it is worth driving, exercise a feature the way a user does, keep evidence, and bug bash it. Writes the skill, a feature map with one explore charter per feature, and proves it once. Use for /create-verification-skill, "make a verification skill for this repo", or when a project has no scripted way to prove UI behavior.

★ 1.6k
hashgraph-online
hashgraph-online/awesome-codex-plugins

deja-search

deja-vu memory — search the user's past AI coding sessions with the deja CLI. Holds what they already settled and never wrote into the repo (naming rules, banned or already-tried tools, required checks before merge, preferred values, who owns what). Use before your first edit in a task, before you call a change done or ready to merge, before you recommend a tool, library, design or reviewer, before re-debugging an error that may already be solved, when they say 'didn't we fix this before', 'what did we decide about X' or 'I already have that', and before telling them something on this machine does not exist.

★ 1.6k
hashgraph-online
hashgraph-online/awesome-codex-plugins

deja-reflect

End-of-session reflection and memory curation with deja-vu. Review what was decided, what commands broke and what repaired them, and record durable project rules or decisions with deja remember so the next session starts with them.

★ 1.6k
hashgraph-online
hashgraph-online/awesome-codex-plugins

revid

Make, edit, export and post AI videos with Revid (revid.ai) through the Revid MCP server. Use whenever the user wants a short-form or long-form video made (from an idea, a script, a URL, an article, a PDF, a tweet, a product page, their own clips, or a song), wants to find what is going viral on TikTok, YouTube, Instagram or X before making something, wants to edit or re-export an existing Revid project, generate a standalone AI image or clip, make a series episode, or publish and schedule a finished video to connected social accounts. Covers setup (hosted MCP with OAuth, or an API key), the 36 tools, credits, polling and the rules for publishing. Routes to the more specific revid-* skills when one fits.

图像与多媒体Skill与智能体开发
★ 1.6k