@culvii/cli 0.0.7 → 0.0.9

package/skills/culvii/references/cli-reference.md

@@ -0,0 +1,217 @@
+# CLI reference
+
+Generated from the registered Commander command tree. Disabled and unregistered commands are excluded.
+
+## `culvii`
+
+Culvii CLI
+
+Options:
+
+- `--debug` — Enable debug logging to ~/.culvii/logs/cli.log
+- `-V, --version` — output the version number
+
+## `culvii auth`
+
+Authentication commands
+
+## `culvii auth login`
+
+Login to Culvii
+
+## `culvii auth logout`
+
+Logout of Culvii
+
+## `culvii deploy`
+
+Deploy an agent
+
+Options:
+
+- `--env <env>` (required) — Environment (sandbox or prod)
+- `--workspace-slug <slug>` — Target workspace slug (overrides config)
+
+## `culvii dev`
+
+Start the development server and file watcher
+
+## `culvii env [environment]`
+
+Show or switch the active environment (dev, sandbox, prod)
+
+## `culvii execution`
+
+Inspect workflow executions
+
+Options:
+
+- `--env <env>` — Target environment (dev, sandbox, or prod)
+
+## `culvii execution get <execution-id>`
+
+Get execution details, step outputs, and errors
+
+Options:
+
+- `--json` — Print machine-readable JSON
+
+## `culvii execution list`
+
+List workspace executions
+
+Options:
+
+- `--json` — Print machine-readable JSON
+- `--limit <limit>` — Results per page
+- `--page <page>` — Page number
+- `--status <status>` — Filter by execution status
+- `--trigger <trigger>` — Filter by trigger type
+- `--workflow <workflow>` — Filter by workflow reference key or ID
+
+## `culvii init [project-name]`
+
+Scaffold a new Culvii project with working agent and workflow files
+
+Options:
+
+- `--here` — Scaffold into the current directory instead of a named subdirectory
+- `--provider <provider>` — Model provider: openai | anthropic | gemini | google-vertex
+- `--template <template>` — Template: agent-only | agent-workflow | agent-workflow-tool | multi-agent
+- `-y, --yes` — Accept all defaults non-interactively
+
+## `culvii integration`
+
+List and manage Culvii integrations
+
+Options:
+
+- `--env <env>` — Target environment (dev, sandbox, or prod)
+
+## `culvii integration list`
+
+List available integration credentials and their scopes
+
+Options:
+
+- `--json` — Print machine-readable JSON
+
+## `culvii login`
+
+Login to Culvii
+
+## `culvii logout`
+
+Logout of Culvii
+
+## `culvii rename <old-slug> <new-slug>`
+
+Rename a resource
+
+Options:
+
+- `--env <env>` — Target environment (sandbox or prod)
+- `--resource-type <type>` (required) — Type of resource (agent)
+- `--workspace-slug <slug>` — Target workspace slug (overrides config)
+
+## `culvii skill`
+
+Manage AI agent skills
+
+## `culvii skill install`
+
+Install the Culvii authoring skill bundle for AI agents
+
+Options:
+
+- `--agent <agent>` — Specify agent (claude|cursor|windsurf|github|opencode|gemini|kiro|codex|agents)
+- `--force` — Replace an existing unmanaged Culvii skill directory
+- `--global` — Install globally instead of in the current project
+
+## `culvii surface`
+
+Surface authoring and review commands
+
+Options:
+
+- `--env <env>` — Target environment (dev, sandbox, or prod)
+
+## `culvii surface review`
+
+Review custom runtime surfaces
+
+## `culvii surface review create`
+
+Create a browser review for selected execution surfaces
+
+Options:
+
+- `--file <path>` (required) — JSON file with { items: [{ executionId, nodeId }] }
+- `--json` — Print machine-readable JSON
+
+## `culvii surface review get <review-id>`
+
+Get selected surfaces and the user feedback collected so far
+
+Options:
+
+- `--json` — Print machine-readable JSON
+
+## `culvii surface review open <review-id>`
+
+Open a surface review in the Culvii Console
+
+## `culvii switch [tenant-slug]`
+
+Switch the active tenant context
+
+## `culvii whoami`
+
+Show current authenticated user context
+
+Options:
+
+- `--json` — Output as JSON
+
+## `culvii workflow`
+
+Workflow execution commands
+
+Options:
+
+- `--env <env>` — Target environment (dev, sandbox, or prod)
+
+## `culvii workflow run <workflow>`
+
+Start a workflow execution
+
+Options:
+
+- `--json` — Print machine-readable JSON
+- `--mocks-file <path>` — JSON object mapping node IDs to mock outputs
+- `--payload <json>` — Trigger payload as JSON
+- `--payload-file <path>` — Read trigger payload from a JSON file
+- `--trigger <node-name>` (required) — Trigger node name
+
+## `culvii workspace`
+
+Workspace commands
+
+Options:
+
+- `--env <env>` — Target environment (sandbox or prod)
+
+## `culvii workspace create`
+
+Create a new workspace
+
+Options:
+
+- `--name <name>` (required) — Workspace name
+- `--slug <slug>` (required) — Workspace slug
+
+## `culvii workspace list`
+
+List workspaces
+
+Further reading: [`culvii dev`](https://culvii.com/docs/cli/dev/) and [`culvii deploy`](https://culvii.com/docs/cli/deploy/).

package/skills/culvii/references/feedback-cycle.md

@@ -0,0 +1,43 @@
+# Workflow feedback cycle
+
+Use this only after `culvii dev` has validated and synced the requested draft resources.
+
+1. Keep the tracked `culvii dev` process running.
+2. Execute feedback only through a `manualTrigger`. If the selected authored trigger is cron, webhook, or polling, first snapshot its exact type, params, credentials, and other trigger-specific fields. Temporarily change that same step to `manualTrigger` while preserving its `id`, `name`, connections, and attached surfaces; remove incompatible trigger-only fields until restoration. Let the active `culvii dev` watcher validate and sync the temporary graph before continuing.
+3. Create a JSON payload matching the real runtime trigger's expected output shape. This representative payload becomes the mocked manual-trigger output automatically and supplies both downstream workflow data and surface preview data. Never mock or invoke the cron, webhook, or polling implementation itself.
+4. For every node reachable from the manual trigger whose generated node reference says `Feedback mock: required`, create a mock output keyed by the node's authored `id`. Do not include cron, webhook, or polling triggers in the mock file. Do not mock `core.agent`; agents must execute through their workflow node.
+5. Start the draft workflow with `culvii workflow run <workflow-reference-key> --trigger <manual-node-name> --payload-file <path> --mocks-file <path> --json`.
+6. Fetch `culvii execution get <execution-id> --json` until its status is `success` or `failed`. This status polling is not a workflow wait and must not resume paused nodes.
+7. On failure, inspect the execution error and the failing step's input/output ports. Fix the authored resource or mock fixture, allow `culvii dev` to resync, and rerun. Stop after three failed attempts and report the remaining error unless the user asks to continue.
+8. On success, inspect the final and intermediate step outputs for the requested behavior. If the workflow has custom candidate surfaces, use these outputs to complete [surface review](surface-feedback.md#validate-execute-and-review). Return to the development lifecycle only after all surface decisions are resolved; keep `culvii dev` running until then.
+9. In success, failure, cancellation, and blocker cleanup, restore the temporarily converted trigger exactly from the snapshot. Wait for `culvii dev` to validate and sync the restored authored graph before stopping the dev process. Do not modify an originally authored manual trigger.
+
+The platform applies these execution-only overrides; the agent must not write them into the authored files:
+
+- Every node receives `pauseAfterExecution: false`.
+- `core.agent` receives `autoCloseSession: true`.
+- Agents outside the workflow are not executed.
+- Surfaces do not control execution progress. Their visual approval uses sanitized execution output but does not change workflow execution.
+
+## Mock file
+
+The file is a JSON object keyed by workflow node ID. An object produces one output item; an array produces multiple items on the first output. Use `batches` only when a node needs multiple output indexes.
+
+```json
+{
+  "gmail-send": { "id": "mock-message-1", "threadId": "mock-thread-1" },
+  "outlook-read": [
+    { "id": "mail-1", "subject": "Urgent request", "from": "person@example.com" }
+  ],
+  "multi-output-node": {
+    "batches": [
+      [{ "matched": true }],
+      [{ "matched": false }]
+    ]
+  }
+}
+```
+
+In mock mode, the trigger payload becomes the selected manual trigger's output. Feedback execution rejects non-manual triggers. A node marked as requiring a feedback mock fails safely when its mock is missing instead of executing. Nodes not marked this way, including core workflow logic and workflow-contained agents, execute normally.
+
+Use `culvii execution list --workflow <workflow> --page <n> --limit <n> --json` to find earlier runs. Use [cli-reference.md](cli-reference.md) for the current generated flags.

package/skills/culvii/references/integrations.md

@@ -0,0 +1,13 @@
+# Integration selection
+
+Use this flow whenever an agent model or workflow node requires a Culvii credential ID.
+
+1. After authentication and tenant selection, run `culvii integration list --json` in the user's project.
+2. Use only a connected integration whose provider, platform scope, and OAuth scopes cover the requested operation. The command returns each usable `credentialId` and the active tenant/environment's `addUrl`.
+3. If exactly one integration matches, use its `credentialId` directly in the authored resource without asking the user to copy it.
+4. If several match, ask one in-run selectable question. Label each option with its integration name, provider, and scope; retain the option-to-`credentialId` mapping internally.
+5. If none match, show the exact `addUrl` returned by the command on its own line as `🟨 **Integration URL:** \`<addUrl>\`` and ask the user to add or connect the integration. The yellow marker is portable; do not emit raw ANSI color codes because structured question dialogs control their own text colors. Keep the current agent run open. After the user confirms completion, rerun `culvii integration list --json` and continue from its real output.
+
+Never ask the user to find, copy, or paste a credential ID. Never invent one. A tenant-scoped integration applies tenant-wide; an environment-scoped integration applies only to its environment; a workspace-scoped integration applies only to the listed workspace IDs. Treat `connected: false` or insufficient OAuth scopes as unusable and explain what must be connected or authorized at the displayed URL.
+
+Users can run `culvii integration list` themselves for a readable table or add `--json` for machine-readable output. See [the generated CLI reference](cli-reference.md) for current flags.

package/skills/culvii/references/kit-api.md

@@ -0,0 +1,35 @@
+# Kit API
+
+Generated from the public `@culvii/kit` source exports. Import authoring APIs only from `@culvii/kit`.
+
+## Key signatures
+
+- `new MultiAgentEngine(primaryConfig: PrimaryAgentConfig)`
+- `createAgentModelConfigFromRef(modelRef, credentialId, options?)`
+- `defineWorkflowTool({ name, description, workflowId, triggerNodeName?, inputSchema? })`
+- `new Workflow({ id, name, active?, steps? })`
+- `new Step({ id?, name, type, params?, credentials?, disabled?, onError?, ui? })`
+- `step.connectTo(nextStep, { fromPort?, fromOutputIndex?, toInputIndex? })`
+- `new Surface({ id, name, description?, slot, definition })`
+
+## Agents
+
+`AgentBaseConfig`, `AgentIntegrationConfig`, `AgentMetadata`, `AgentModel`, `AgentModelConfig`, `AgentModelConfigOptions`, `AgentModelRef`, `AgentPermissions`, `AgentRole`, `AgentRuntimeDetails`, `BuiltInModelProvider`, `MultiAgentEngine`, `PrimaryAgentConfig`, `SANDBOX_TOOL_NAMES`, `SandboxToolName`, `SandboxToolPermission`, `SecondaryAgentConfig`, `SecondaryAgentMetadata`, `Tool`, `WorkflowToolConfig`, `createAgentModelConfigFromRef`, `defineWorkflowTool`, `parseAgentModelRef`
+
+## Development registry
+
+`DevRegistry`
+
+## Shared types
+
+`CulviiConfig`, `CulviiRuntime`, `DIDDocument`, `Guard`, `HealthReport`, `HookContext`, `IdentityConfig`, `IdentityRef`, `MaybePromise`, `MemoryConfig`, `MemoryType`, `Message`, `ObserveConfig`, `Permission`, `Plan`, `Policy`, `RetryConfig`, `SandboxConfig`, `Schema`, `Skill`, `StepTrace`, `Tool`, `ToolCallEvent`, `ToolResultEvent`, `Usage`
+
+## Surfaces
+
+`BinaryDescriptor`, `DefaultSurfaceSelection`, `ExecutionUiBinaryNodeValues`, `ExecutionUiDataModel`, `ExecutionUiNodeValues`, `ExecutionUiSurfaceEntry`, `ExecutionUiSurfaceMap`, `NodeSurfaceSlot`, `RuntimeSurface`, `RuntimeSurfaceAction`, `RuntimeSurfaceComponent`, `SerializedNodeUiConfig`, `SerializedSurface`, `SerializedSurfaceSelection`, `Surface`, `SurfaceAction`, `SurfaceBinding`, `SurfaceComponent`, `SurfaceConfig`, `SurfaceFormField`, `SurfaceReference`, `UISurfaceDefinition`, `UIValue`, `isSurface`
+
+## Workflows and steps
+
+`ConnectStepOptions`, `CredentialedNodeType`, `LoopBodyBuilder`, `NodeExpression`, `NodeParams`, `NodeParamsBase`, `NodeType`, `NodeTypeInput`, `NodeTypeParamsMap`, `NodeUiConfig`, `RegisteredNodeType`, `Resolvable`, `SerializedWorkflow`, `SerializedWorkflowStep`, `Step`, `StepCredentials`, `StepSurfaceSelection`, `Workflow`, `WorkflowConfig`, `WorkflowLoopConfig`, `WorkflowRuntimeConnection`, `WorkflowRuntimeConnections`, `WorkflowRuntimeNode`, `WorkflowRuntimeShape`, `WorkflowStepConfig`, `WorkflowStepConnection`
+
+Further reading: [Workflow API](https://culvii.com/docs/sdk-reference/api-reference/workflow-api/), [Multi-Agent API](https://culvii.com/docs/sdk-reference/api-reference/multi-agent-api/), and [Authentication](https://culvii.com/docs/sdk-reference/guides/authentication/).

package/skills/culvii/references/node-argocd.md

@@ -0,0 +1,44 @@
+# argocd
+
+- Type ID: `argocd`
+- Kind: `action`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `argocd`
+- Feedback mock: required
+
+Resolve the credential through [integration selection](integrations.md); never ask the user to copy a credential ID.
+
+## Props
+
+- `resource`: options; required
+  - Values: `application`, `project`
+- `operation`: options; required; when resource=application
+  - Values: `getApplication`, `syncApplication`, `listApplications`, `getResourceTree`
+- `operation`: options; required; when resource=project
+  - Values: `listProjects`
+- `applicationName`: string; required; when resource=application, operation=getApplication
+- `refresh`: options; optional; when resource=application, operation=getApplication
+  - Values: ``, `normal`, `hard`
+- `applicationName`: string; required; when resource=application, operation=syncApplication
+- `revision`: string; optional; when resource=application, operation=syncApplication
+- `prune`: boolean; optional; when resource=application, operation=syncApplication
+- `dryRun`: boolean; optional; when resource=application, operation=syncApplication
+- `strategy`: options; optional; when resource=application, operation=syncApplication
+  - Values: `hook`, `apply`
+- `force`: boolean; optional; when resource=application, operation=syncApplication, strategy=apply
+- `project`: string; optional; when resource=application, operation=listApplications
+- `applicationName`: string; required; when resource=application, operation=getResourceTree
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'argocd',
+  name: 'argocd',
+  type: 'argocd',
+  credentials: { name: 'argocd', id: 'credential-id' },
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii argocd node](https://culvii.com/docs/nodes/argocd/).

package/skills/culvii/references/node-clickup-doc.md

@@ -0,0 +1,38 @@
+# clickupDoc
+
+- Type ID: `clickupDoc`
+- Kind: `action`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `clickup`
+- Feedback mock: required
+
+Resolve the credential through [integration selection](integrations.md); never ask the user to copy a credential ID.
+
+## Props
+
+- `resource`: options; required
+  - Values: `doc`
+- `operation`: options; required; when resource=doc
+  - Values: `create`, `read`
+- `workspaceId`: string; required; when resource=doc, operation=create
+- `name`: string; required; when resource=doc, operation=create
+- `parentType`: options; required; when resource=doc, operation=create
+  - Values: `12`, `4`, `5`, `6`
+- `parentId`: string; optional; when resource=doc, operation=create
+- `content`: string; optional; when resource=doc, operation=create
+- `workspaceId`: string; required; when resource=doc, operation=read
+- `docId`: string; required; when resource=doc, operation=read
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'clickup-doc',
+  name: 'clickupDoc',
+  type: 'clickupDoc',
+  credentials: { name: 'clickup', id: 'credential-id' },
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii clickupDoc node](https://culvii.com/docs/nodes/clickup-doc/).

package/skills/culvii/references/node-clickup-ticket.md

@@ -0,0 +1,47 @@
+# clickupTicket
+
+- Type ID: `clickupTicket`
+- Kind: `action`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `clickup`
+- Feedback mock: required
+
+Resolve the credential through [integration selection](integrations.md); never ask the user to copy a credential ID.
+
+## Props
+
+- `resource`: options; required
+  - Values: `ticket`
+- `operation`: options; required; when resource=ticket
+  - Values: `create`, `update`, `updateStatus`, `updateDescription`
+- `listId`: string; required; when resource=ticket, operation=create
+- `name`: string; required; when resource=ticket, operation=create
+- `description`: string; optional; when resource=ticket, operation=create
+- `status`: string; optional; when resource=ticket, operation=create
+- `assignees`: string; optional; when resource=ticket, operation=create
+- `priority`: options; optional; when resource=ticket, operation=create
+  - Values: ``, `1`, `2`, `3`, `4`
+- `taskId`: string; required; when resource=ticket, operation=update
+- `name`: string; optional; when resource=ticket, operation=update
+- `description`: string; optional; when resource=ticket, operation=update
+- `status`: string; optional; when resource=ticket, operation=update
+- `priority`: options; optional; when resource=ticket, operation=update
+  - Values: ``, `1`, `2`, `3`, `4`
+- `taskId`: string; required; when resource=ticket, operation=updateStatus
+- `status`: string; required; when resource=ticket, operation=updateStatus
+- `taskId`: string; required; when resource=ticket, operation=updateDescription
+- `description`: string; required; when resource=ticket, operation=updateDescription
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'clickup-ticket',
+  name: 'clickupTicket',
+  type: 'clickupTicket',
+  credentials: { name: 'clickup', id: 'credential-id' },
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii clickupTicket node](https://culvii.com/docs/nodes/clickup-ticket/).

package/skills/culvii/references/node-core-agent.md

@@ -0,0 +1,27 @@
+# core.agent
+
+- Type ID: `core.agent`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: not required
+
+## Props
+
+- `agentReferenceKey`: string; required
+- `prompt`: string; required
+- `autoCloseSession`: boolean; optional
+- `outputFormat`: json; optional
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-agent',
+  name: 'core.agent',
+  type: 'core.agent',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.agent node](https://culvii.com/docs/nodes/core/agent/).

package/skills/culvii/references/node-core-browser-automation.md

@@ -0,0 +1,53 @@
+# core.browserAutomation
+
+- Type ID: `core.browserAutomation`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: required
+
+## Props
+
+- `url`: string; required
+- `timeoutMs`: number; optional
+- `includeHtml`: boolean; optional
+- `auth`: object; required
+  - `type`: options; required
+    - Values: `none`, `form`
+  - `loginUrl`: string; optional; when type=form
+  - `usernameSelector`: string; optional; when type=form
+  - `passwordSelector`: string; optional; when type=form
+  - `submitSelector`: string; optional; when type=form
+  - `otpSelector`: string; optional; when type=form
+  - `otpSubmitSelector`: string; optional; when type=form
+  - `useTotp`: boolean; optional; when type=form
+- `steps`: array; optional
+  - `step`: object; optional
+    - `type`: options; optional
+      - Values: `goto`, `fill`, `click`, `waitForSelector`, `waitForTimeout`, `extract`
+    - `url`: string; optional; when type=goto
+    - `waitUntil`: options; optional; when type=goto
+      - Values: `load`, `domcontentloaded`, `networkidle`
+    - `selector`: string; optional; when type=fill|click|waitForSelector
+    - `value`: string; optional; when type=fill
+    - `timeoutMs`: number; optional; when type=waitForSelector|waitForTimeout
+    - `selectors`: array; optional; when type=extract
+      - `selectorRule`: object; optional
+        - `key`: string; optional
+        - `selector`: string; optional
+        - `type`: options; optional
+          - Values: `text`, `texts`, `attr`, `attrs`
+        - `attrName`: string; optional; when type=attr|attrs
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-browser-automation',
+  name: 'core.browserAutomation',
+  type: 'core.browserAutomation',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.browserAutomation node](https://culvii.com/docs/nodes/core/browser-automation/).

package/skills/culvii/references/node-core-code.md

@@ -0,0 +1,28 @@
+# core.code
+
+- Type ID: `core.code`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: not required
+
+## Props
+
+- `code`: string; required
+- `codeType`: options; optional
+  - Values: `js`, `py`
+- `codeParams`: object; optional
+- `globalParams`: object; optional
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-code',
+  name: 'core.code',
+  type: 'core.code',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.code node](https://culvii.com/docs/nodes/core/code/).

package/skills/culvii/references/node-core-http.md

@@ -0,0 +1,42 @@
+# core.http
+
+- Type ID: `core.http`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: required
+
+## Props
+
+- `method`: options; required
+  - Values: `GET`, `POST`, `PUT`, `PATCH`, `DELETE`, `HEAD`, `OPTIONS`
+- `url`: string; required
+- `authentication`: options; optional
+  - Values: `none`, `bearerToken`, `basicAuth`
+- `bearerToken`: string; optional; when authentication=bearerToken
+- `username`: string; optional; when authentication=basicAuth
+- `password`: string; optional; when authentication=basicAuth
+- `queryParameters`: object; optional
+- `headers`: object; optional
+- `bodyContentType`: options; optional
+  - Values: `json`, `form`, `form-urlencoded`, `raw`
+- `body`: string; optional
+- `responseFormat`: options; optional
+  - Values: `json`, `text`, `autodetect`
+- `options`: object; optional
+  - `timeout`: number; optional
+  - `followRedirects`: boolean; optional
+  - `ignoreSSL`: boolean; optional
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-http',
+  name: 'core.http',
+  type: 'core.http',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.http node](https://culvii.com/docs/nodes/core/http/).

package/skills/culvii/references/node-core-if.md

@@ -0,0 +1,26 @@
+# core.if
+
+- Type ID: `core.if`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: not required
+
+## Props
+
+- `conditions`: array; required
+- `combineOperation`: options; required
+  - Values: `all`, `any`
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-if',
+  name: 'core.if',
+  type: 'core.if',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.if node](https://culvii.com/docs/nodes/core/if/).

package/skills/culvii/references/node-core-jwt.md

@@ -0,0 +1,31 @@
+# core.jwt
+
+- Type ID: `core.jwt`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: not required
+
+## Props
+
+- `operation`: options; required
+  - Values: `sign`, `verify`
+- `payload`: object; optional; when operation=sign
+- `token`: string; optional; when operation=verify
+- `secret`: string; required
+- `expiresIn`: string; optional; when operation=sign
+- `issuer`: string; optional
+- `audience`: string; optional
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-jwt',
+  name: 'core.jwt',
+  type: 'core.jwt',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.jwt node](https://culvii.com/docs/nodes/core/jwt/).

package/skills/culvii/references/node-core-loop-break.md

@@ -0,0 +1,26 @@
+# core.loopBreak
+
+- Type ID: `core.loopBreak`
+- Kind: `action`
+- Credentials: not required
+- Feedback mock: not required
+
+## Props
+
+- `conditions`: array; required
+- `combineOperation`: options; required
+  - Values: `all`, `any`
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'core-loop-break',
+  name: 'core.loopBreak',
+  type: 'core.loopBreak',
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii core.loopBreak node](https://culvii.com/docs/nodes/core/loop-break/).