@culvii/cli 0.0.8 → 0.1.0

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

@@ -2,7 +2,7 @@
 
 - Type ID: `slack`
 - Kind: `action`
-- Credentials: required as `credentials: { name, id }`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `slack_bot`, `slack_user`
 - Feedback mock: required
 
 Resolve the credential through [integration selection](integrations.md); never ask the user to copy a credential ID.
@@ -10,17 +10,141 @@ Resolve the credential through [integration selection](integrations.md); never a
 ## Props
 
 - `resource`: options; required
-  - Values: `message`
+  - Values: `message`, `reaction`, `user`, `channel`, `file`, `search`, `interaction`
 - `operation`: options; required; when resource=message
-  - Values: `send`
+  - Values: `send`, `sendByEmail`, `update`, `delete`, `getReplies`, `getPermalink`, `schedule`, `listScheduled`, `cancelScheduled`, `sendEphemeral`
 - `channel`: string; required; when resource=message, operation=send
 - `text`: string; required; when resource=message, operation=send
 - `additionalFields`: object; optional; when resource=message, operation=send
   - `thread_ts`: string; optional
+  - `reply_broadcast`: boolean; optional
+  - `blocks`: json; optional
   - `mrkdwn`: boolean; optional
   - `link_names`: boolean; optional
   - `unfurl_links`: boolean; optional
   - `unfurl_media`: boolean; optional
+- `email`: string; required; when resource=message, operation=sendByEmail
+- `text`: string; required; when resource=message, operation=sendByEmail
+- `messageOptions`: object; optional; when resource=message, operation=sendByEmail
+  - `blocks`: json; optional
+  - `mrkdwn`: boolean; optional
+  - `link_names`: boolean; optional
+  - `unfurl_links`: boolean; optional
+  - `unfurl_media`: boolean; optional
+- `channel`: string; required; when resource=message, operation=sendEphemeral
+- `userId`: string; required; when resource=message, operation=sendEphemeral
+- `text`: string; required; when resource=message, operation=sendEphemeral
+- `ephemeralOptions`: object; optional; when resource=message, operation=sendEphemeral
+  - `thread_ts`: string; optional
+  - `blocks`: json; optional
+  - `link_names`: boolean; optional
+- `channel`: string; required; when resource=message, operation=schedule
+- `text`: string; required; when resource=message, operation=schedule
+- `postAt`: dateTime; required; when resource=message, operation=schedule
+- `scheduleOptions`: object; optional; when resource=message, operation=schedule
+  - `thread_ts`: string; optional
+  - `reply_broadcast`: boolean; optional
+  - `blocks`: json; optional
+  - `link_names`: boolean; optional
+  - `unfurl_links`: boolean; optional
+  - `unfurl_media`: boolean; optional
+- `channelId`: string; optional; when resource=message, operation=listScheduled
+- `oldest`: dateTime; optional; when resource=message, operation=listScheduled
+- `latest`: dateTime; optional; when resource=message, operation=listScheduled
+- `limit`: number; optional; when resource=message, operation=listScheduled
+- `cursor`: string; optional; when resource=message, operation=listScheduled
+- `channelId`: string; required; when resource=message, operation=cancelScheduled
+- `scheduledMessageId`: string; required; when resource=message, operation=cancelScheduled
+- `channelId`: string; required; when resource=message, operation=update
+- `messageTs`: string; required; when resource=message, operation=update
+- `text`: string; required; when resource=message, operation=update
+- `blocks`: json; optional; when resource=message, operation=update
+- `channelId`: string; required; when resource=message, operation=delete
+- `messageTs`: string; required; when resource=message, operation=delete
+- `channelId`: string; required; when resource=message, operation=getReplies
+- `threadTs`: string; required; when resource=message, operation=getReplies
+- `limit`: number; optional; when resource=message, operation=getReplies
+- `output`: options; optional; when resource=message, operation=getReplies
+  - Values: `simple`, `raw`
+- `cursor`: string; optional; when resource=message, operation=getReplies
+- `channelId`: string; required; when resource=message, operation=getPermalink
+- `messageTs`: string; required; when resource=message, operation=getPermalink
+- `operation`: options; required; when resource=reaction
+  - Values: `add`, `remove`
+- `channelId`: string; required; when resource=reaction, operation=add
+- `messageTs`: string; required; when resource=reaction, operation=add
+- `emoji`: string; required; when resource=reaction, operation=add
+- `channelId`: string; required; when resource=reaction, operation=remove
+- `messageTs`: string; required; when resource=reaction, operation=remove
+- `emoji`: string; required; when resource=reaction, operation=remove
+- `operation`: options; required; when resource=user
+  - Values: `getByEmail`, `getMany`, `getProfile`, `updateProfile`
+- `email`: string; required; when resource=user, operation=getByEmail
+- `output`: options; optional; when resource=user, operation=getByEmail
+  - Values: `simple`, `raw`
+- `limit`: number; optional; when resource=user, operation=getMany
+- `output`: options; optional; when resource=user, operation=getMany
+  - Values: `simple`, `raw`
+- `cursor`: string; optional; when resource=user, operation=getMany
+- `user`: string; optional; when resource=user, operation=getProfile
+- `user`: string; optional; when resource=user, operation=updateProfile
+- `clearStatus`: boolean; optional; when resource=user, operation=updateProfile
+- `status_text`: string; optional; when resource=user, operation=updateProfile
+- `status_emoji`: string; optional; when resource=user, operation=updateProfile
+- `status_expiration`: string; optional; when resource=user, operation=updateProfile
+- `first_name`: string; optional; when resource=user, operation=updateProfile
+- `last_name`: string; optional; when resource=user, operation=updateProfile
+- `email`: string; optional; when resource=user, operation=updateProfile
+- `operation`: options; required; when resource=channel
+  - Values: `getMany`, `create`, `archive`, `setTopic`, `invite`, `remove`
+- `channelTypes`: options; optional; when resource=channel, operation=getMany
+  - Values: `public`, `private`, `publicAndPrivate`, `directMessages`, `groupDirectMessages`, `all`
+- `excludeArchived`: boolean; optional; when resource=channel, operation=getMany
+- `limit`: number; optional; when resource=channel, operation=getMany
+- `output`: options; optional; when resource=channel, operation=getMany
+  - Values: `simple`, `raw`
+- `cursor`: string; optional; when resource=channel, operation=getMany
+- `channelName`: string; required; when resource=channel, operation=create
+- `isPrivate`: boolean; optional; when resource=channel, operation=create
+- `channelId`: string; required; when resource=channel, operation=archive
+- `channelId`: string; required; when resource=channel, operation=setTopic
+- `topic`: string; required; when resource=channel, operation=setTopic
+- `channelId`: string; required; when resource=channel, operation=invite
+- `userIds`: string; required; when resource=channel, operation=invite
+- `channelId`: string; required; when resource=channel, operation=remove
+- `userId`: string; required; when resource=channel, operation=remove
+- `operation`: options; required; when resource=file
+  - Values: `upload`, `getInfo`, `download`
+- `binaryPropertyName`: string; required; when resource=file, operation=upload
+- `channelId`: string; optional; when resource=file, operation=upload
+- `uploadOptions`: object; optional; when resource=file, operation=upload
+  - `filename`: string; optional
+  - `title`: string; optional
+  - `initial_comment`: string; optional
+  - `thread_ts`: string; optional
+  - `alt_txt`: string; optional
+- `fileId`: string; required; when resource=file, operation=getInfo
+- `output`: options; optional; when resource=file, operation=getInfo
+  - Values: `simple`, `raw`
+- `fileId`: string; required; when resource=file, operation=download
+- `binaryPropertyName`: string; required; when resource=file, operation=download
+- `operation`: options; required; when resource=search
+  - Values: `messages`, `files`
+- `query`: string; required; when resource=search
+- `count`: number; optional; when resource=search
+- `page`: number; optional; when resource=search
+- `sort`: options; optional; when resource=search
+  - Values: `score`, `timestamp`
+- `sort_dir`: options; optional; when resource=search
+  - Values: `asc`, `desc`
+- `operation`: options; required; when resource=interaction
+  - Values: `respond`
+- `responseUrl`: string; required; when resource=interaction, operation=respond
+- `text`: string; optional; when resource=interaction, operation=respond
+- `blocks`: json; optional; when resource=interaction, operation=respond
+- `replaceOriginal`: boolean; optional; when resource=interaction, operation=respond
+- `responseType`: options; optional; when resource=interaction, operation=respond
+  - Values: `in_channel`, `ephemeral`
 - `pauseAfterExecution`: boolean; optional
 
 ## Example
@@ -30,7 +154,7 @@ new Step({
   id: 'slack',
   name: 'slack',
   type: 'slack',
-  credentials: { name: 'Stored credential', id: 'credential-id' },
+  credentials: { name: 'slack_bot', id: 'credential-id' },
   params: { /* use the required props and conditional values above */ },
 });
 ```

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

@@ -0,0 +1,61 @@
+# supabase
+
+- Type ID: `supabase`
+- Kind: `action`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `supabase`
+- 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: `database`, `storage`
+- `operation`: options; required; when resource=database
+  - Values: `getRows`, `insertRow`, `updateRows`, `upsertRow`, `deleteRows`, `callFunction`
+- `operation`: options; required; when resource=storage
+  - Values: `uploadFile`, `downloadFile`, `listFiles`, `deleteFile`
+- `table`: string; required; when resource=database, operation=getRows
+- `columns`: string; optional; when resource=database, operation=getRows
+- `filters`: json; optional; when resource=database, operation=getRows
+- `order`: string; optional; when resource=database, operation=getRows
+- `limit`: number; optional; when resource=database, operation=getRows
+- `table`: string; required; when resource=database, operation=insertRow
+- `rowData`: json; required; when resource=database, operation=insertRow
+- `table`: string; required; when resource=database, operation=updateRows
+- `filters`: json; required; when resource=database, operation=updateRows
+- `rowData`: json; required; when resource=database, operation=updateRows
+- `table`: string; required; when resource=database, operation=upsertRow
+- `rowData`: json; required; when resource=database, operation=upsertRow
+- `onConflict`: string; optional; when resource=database, operation=upsertRow
+- `table`: string; required; when resource=database, operation=deleteRows
+- `filters`: json; required; when resource=database, operation=deleteRows
+- `functionName`: string; required; when resource=database, operation=callFunction
+- `args`: json; optional; when resource=database, operation=callFunction
+- `bucket`: string; required; when resource=storage, operation=uploadFile
+- `filePath`: string; required; when resource=storage, operation=uploadFile
+- `binaryPropertyName`: string; required; when resource=storage, operation=uploadFile
+- `upsert`: boolean; optional; when resource=storage, operation=uploadFile
+- `bucket`: string; required; when resource=storage, operation=downloadFile
+- `filePath`: string; required; when resource=storage, operation=downloadFile
+- `binaryPropertyName`: string; required; when resource=storage, operation=downloadFile
+- `bucket`: string; required; when resource=storage, operation=listFiles
+- `folderPath`: string; optional; when resource=storage, operation=listFiles
+- `limit`: number; optional; when resource=storage, operation=listFiles
+- `bucket`: string; required; when resource=storage, operation=deleteFile
+- `filePath`: string; required; when resource=storage, operation=deleteFile
+- `pauseAfterExecution`: boolean; optional
+
+## Example
+
+```ts
+new Step({
+  id: 'supabase',
+  name: 'supabase',
+  type: 'supabase',
+  credentials: { name: 'supabase', id: 'credential-id' },
+  params: { /* use the required props and conditional values above */ },
+});
+```
+
+Further reading: [Culvii supabase node](https://culvii.com/docs/nodes/supabase/).

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

@@ -2,7 +2,7 @@
 
 - Type ID: `twilio`
 - Kind: `action`
-- Credentials: required as `credentials: { name, id }`
+- Credentials: required as `credentials: { name, id }`; `name` must be one of: `twilio`
 - Feedback mock: required
 
 Resolve the credential through [integration selection](integrations.md); never ask the user to copy a credential ID.
@@ -31,7 +31,7 @@ new Step({
   id: 'twilio',
   name: 'twilio',
   type: 'twilio',
-  credentials: { name: 'Stored credential', id: 'credential-id' },
+  credentials: { name: 'twilio', id: 'credential-id' },
   params: { /* use the required props and conditional values above */ },
 });
 ```

package/skills/culvii/references/nodes-index.md

@@ -2,26 +2,50 @@
 
 Generated from `GENERATED_NODE_DEFINITIONS`. Open the exact node page before authoring its params.
 
+- [`argocd`](node-argocd.md) — action; credentials required; feedback mock required
+- [`clickupDoc`](node-clickup-doc.md) — action; credentials required; feedback mock required
+- [`clickupTicket`](node-clickup-ticket.md) — action; credentials required; feedback mock required
 - [`core.agent`](node-core-agent.md) — action
 - [`core.browserAutomation`](node-core-browser-automation.md) — action; feedback mock required
 - [`core.code`](node-core-code.md) — action
 - [`core.http`](node-core-http.md) — action; feedback mock required
 - [`core.if`](node-core-if.md) — action
 - [`core.jwt`](node-core-jwt.md) — action
+- [`core.loop`](node-core-loop.md) — action
+- [`core.loopBreak`](node-core-loop-break.md) — action
+- [`core.loopContinue`](node-core-loop-continue.md) — action
+- [`core.loopOutput`](node-core-loop-output.md) — action
+- [`core.merge`](node-core-merge.md) — action
 - [`core.set`](node-core-set.md) — action
+- [`core.switch`](node-core-switch.md) — action
+- [`costExplorer`](node-cost-explorer.md) — action; credentials required; feedback mock required
 - [`cronTrigger`](node-cron-trigger.md) — trigger
 - [`elasticsearch`](node-elasticsearch.md) — action; credentials required; feedback mock required
+- [`github`](node-github.md) — action; credentials required; feedback mock required
+- [`githubPollTrigger`](node-github-poll-trigger.md) — trigger; credentials required; feedback mock required
+- [`githubWebhookTrigger`](node-github-webhook-trigger.md) — trigger; credentials required; feedback mock required
 - [`gmail`](node-gmail.md) — action; credentials required; feedback mock required
 - [`gmailTrigger`](node-gmail-trigger.md) — trigger; credentials required; feedback mock required
 - [`googleCalendar`](node-google-calendar.md) — action; credentials required; feedback mock required
 - [`googleDocs`](node-google-docs.md) — action; credentials required; feedback mock required
+- [`grafanaDashboard`](node-grafana-dashboard.md) — action; credentials required; feedback mock required
+- [`grafanaDataSource`](node-grafana-data-source.md) — action; credentials required; feedback mock required
+- [`grafanaQuery`](node-grafana-query.md) — action; credentials required; feedback mock required
+- [`hibp.listDomains`](node-hibp-list-domains.md) — action; credentials required; feedback mock required
+- [`hibp.stealerLogs`](node-hibp-stealer-logs.md) — action; credentials required; feedback mock required
+- [`kubernetes`](node-kubernetes.md) — action; credentials required; feedback mock required
 - [`manualTrigger`](node-manual-trigger.md) — trigger
 - [`microsoftOutlook`](node-microsoft-outlook.md) — action; credentials required; feedback mock required
 - [`microsoftOutlookTrigger`](node-microsoft-outlook-trigger.md) — trigger; credentials required; feedback mock required
 - [`microsoftSharepoint`](node-microsoft-sharepoint.md) — action; credentials required; feedback mock required
+- [`onionsearch`](node-onionsearch.md) — action; credentials required; feedback mock required
+- [`prometheus`](node-prometheus.md) — action; credentials required; feedback mock required
 - [`s3`](node-s3.md) — action; credentials required; feedback mock required
 - [`slack`](node-slack.md) — action; credentials required; feedback mock required
+- [`slackEventsTrigger`](node-slack-events-trigger.md) — trigger; credentials required; feedback mock required
+- [`slackInteractionsTrigger`](node-slack-interactions-trigger.md) — trigger; credentials required; feedback mock required
 - [`slackTrigger`](node-slack-trigger.md) — trigger; credentials required; feedback mock required
+- [`supabase`](node-supabase.md) — action; credentials required; feedback mock required
 - [`twilio`](node-twilio.md) — action; credentials required; feedback mock required
 - [`waitTimer`](node-wait-timer.md) — action
 - [`webhookTrigger`](node-webhook-trigger.md) — trigger

package/skills/culvii/references/surface-authoring.md

@@ -7,7 +7,18 @@ For workflow-attached surfaces, make selection, validation, preview, and approva
 - Objects: `KeyValue`; wrap important groups in `Card` or `Section`.
 - Arrays: `Table` for comparable records, or repeated domain cards composed with layout components.
 - Metrics: `Stat`, `Money`, `Duration`, or `Timestamp`.
-- Files: `FileChip` bound through `$binary`.
+- Files: `FileChip` bound through `$binary`. Binding `file` alone renders a read-only file
+  summary — it is NOT clickable until you also declare `actions.default` as a `navigate` action
+  whose `href` points at the same descriptor's `downloadUrl`:
+  ```ts
+  {
+    id: 'file', type: 'FileChip',
+    props: { file: { path: '/$binary/Render Report/main/0/0/report' } },
+    actions: {
+      default: { type: 'navigate', href: { path: '/$binary/Render Report/main/0/0/report/downloadUrl' } },
+    },
+  }
+  ```
 - Human decisions: a wait `Form` with input components and `Button`, or `ApproveRejectBar`.
 - Raw/debug values: `JsonInspector` or `Code`.
 

package/skills/culvii/references/surface-display.md

@@ -129,7 +129,7 @@ Props:
 
 ## FileChip
 
-Metadata and download affordance for a produced file.
+Metadata and download affordance for a produced file. Binding `file` alone does not make the chip clickable — declare `actions.default` as a `navigate` action with `href` bound to the same descriptor's `downloadUrl` path (e.g. `{ path: "/$binary/<step>/main/0/0/<key>/downloadUrl" }`), or the chip renders as a read-only file summary with no download button.
 
 - Children: `none`
 - Interaction: `single-action`

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

@@ -1,41 +1,38 @@
 # Surface configuration and visual approval
 
-Use this flow for every newly authored workflow. The agent authors surfaces; the user chooses the experience and approves what they see.
+Use this flow for every newly authored workflow. The agent authors surfaces; the user reviews the exact Culvii renderer before accepting them.
 
 ## Choose before validation
 
-1. After planning the workflow graph and expected node data, ask once through the agent's native in-run question tool whether the user wants to configure runtime widgets or keep the defaults. Do not end the chat run to ask.
-2. If the user chooses configuration, ask for each node in each workflow whether to:
-   - create a custom surface now;
-   - attach a compatible existing surface; or
-   - use the platform default.
-3. Present only real existing surfaces found in project files or command output. Never invent a surface reference. Combine independent node choices into the fewest question-tool prompts the agent supports.
-4. Read [surface authoring](surface-authoring.md) and only the component catalog pages needed for custom choices. Use expected output shape and the workflow's purpose to design the surface; do not ask the user to write component documents.
-5. Create every custom candidate as a real Culvii `Surface` and attach it to its workflow step before the first `culvii dev`. "Candidate" describes its approval state, not a separate resource type. Attach selected existing surfaces now and leave default choices unchanged.
+1. After planning the workflow graph and expected node data, ask once through the agent's native in-run question tool whether the user wants runtime widgets or platform defaults. Do not end the chat run to ask.
+2. If the user chooses configuration, ask for each node in each workflow whether to create a custom surface, attach a real existing surface, or use the default. Combine independent choices into the fewest prompts the agent supports.
+3. Read [surface authoring](surface-authoring.md) and only the component catalog pages needed for custom choices. Use expected output shape and workflow purpose; do not ask the user to write component documents.
+4. Create every custom candidate as a real Culvii `Surface` and attach it to its workflow step before the first `culvii dev`. Attach selected existing surfaces now and leave default choices unchanged.
 
-Do not introduce a wait surface or pause merely to display output. A wait surface is valid only when the requested workflow already requires human input. Surface documents use only supported Culvii components and bindings; the temporary preview HTML described below is never placed in a `.culvii.ts` file.
+Do not add a wait surface or pause merely to display output. A wait surface is valid only when the workflow already needs human input.
 
-## Validate and collect data
+## Validate, execute, and review
 
-Start one tracked `culvii dev` process after the workflow and all candidate surfaces exist. Fix every schema, prop, binding, slot, and cross-resource error reported by the watcher. Never preview an invalid candidate.
+1. Start one tracked `culvii dev` process after the workflow and all candidate surfaces exist. Fix every validation and sync error before previewing.
+2. Run the [workflow feedback cycle](feedback-cycle.md). Its successful execution supplies the preview data. For a cron, webhook, or polling trigger, use its temporary manual-trigger conversion and representative payload; never invoke or mock the runtime trigger implementation.
+3. For every custom surface in that execution, create a review manifest. It includes only the configured nodes, never the full run:
 
-Once the draft graph syncs successfully, run the [workflow feedback cycle](feedback-cycle.md). Use its successful node outputs as preview data. For an authored cron, webhook, or polling trigger, create a representative payload matching its real output shape and deliver it through the temporary manual-trigger conversion. This supplies the trigger's surface and downstream nodes without mocking or invoking the runtime trigger implementation. Sanitize private values and replace credentials, tokens, raw binary, and other secrets with harmless representative values.
+```json
+{
+  "items": [
+    { "executionId": "exc_…", "nodeId": "research-agent" },
+    { "executionId": "exc_…", "nodeId": "report-output" }
+  ]
+}
+```
 
-## Visual approval loop
+4. Run `culvii surface review create --file <manifest.json> --json`. It returns a Console URL. Ask permission before running `culvii surface review open <review-id>`; otherwise show the URL to the user and keep the same chat run open.
+5. The review page uses the production `SurfaceView` renderer, groups items by workflow, and labels each preview as **Mocked output**, **Simulated trigger**, or **Executed output**. It shows only the configured custom surfaces.
+6. Wait for the user to choose **Keep this surface**, **Request changes**, or **Use default** for every item and submit the review. Read `culvii surface review get <review-id> --json` to obtain the decisions; do not ask the user to paste them into chat.
+7. For changes, edit the real surface and let the active `culvii dev` watcher validate/sync it. Create a new review after the next successful feedback execution. For default, remove the custom attachment/resource as appropriate and let `culvii dev` validate it.
 
-For each custom candidate:
-
-1. Create a tracked temporary directory outside the project and generate a self-contained HTML file that approximates the validated Culvii surface with its sanitized execution data. Use inline CSS and only minimal local JavaScript needed for visual interaction. Make no network requests. Escape untrusted text, do not embed credentials or raw binary, and label the page **Visual preview — Culvii runtime rendering may differ**.
-2. Ask for permission before launching the local `file://` page in the user's browser. If the environment cannot open it, provide the local file path and keep the same chat run waiting for review; do not treat the surface as approved.
-3. Ask through the native in-run question tool:
-   - **Keep this widget** — accept the attached custom surface;
-   - **Change it** — collect plain-language feedback in the same run; or
-   - **Use the default** — detach/remove the candidate and keep the platform default.
-4. For **Change it**, edit the real surface, wait for the active `culvii dev` watcher to validate and sync it, regenerate the HTML with the same data, and ask again. Never show a revision that failed validation.
-5. For **Use the default**, remove the custom attachment/resource as appropriate and wait for `culvii dev` to validate the updated graph.
-
-The HTML is a review aid, not a second validator and not a claim of pixel parity with the Culvii runtime. Component contracts and bindings remain authoritative through `culvii dev`.
+The review is an ephemeral, user-scoped record. It resolves the current draft surface and current execution data on each page load, so it never claims that an old preview matches a revised surface. It expires after two hours; recreate it if needed.
 
 ## Finish
 
-Every node must end as accepted custom, selected existing, platform default, or explicitly skipped by the user. After the final workflow execution succeeds, delete only the tracked temporary preview directory and follow the development guide to stop the exact `culvii dev` process. Never report completion while a custom surface is awaiting review.
+Every configured node must end as accepted custom, selected existing, platform default, or explicitly skipped. Restore any temporary trigger conversion, wait for `culvii dev` to sync it, then stop only the exact `culvii dev` process started for this task. Never report completion while a custom surface is awaiting review.