@adminide-stack/yantra-help-browser 12.0.191-alpha.2 → 12.0.192-alpha.0

package/lib/templates/content/chat/cards-and-widgets.md

@@ -0,0 +1,37 @@
+---
+title: 'Cards and Widgets'
+description: 'The rich results that appear inside a conversation: HTML plans, media previews, travel, shopping, events and workflow cards, code diffs and the tool timeline.'
+author: 'Yantra Team'
+---
+
+# Cards and Widgets
+
+Some answers are not paragraphs. Yantra renders them as cards inside the conversation, and most cards open a detail panel or an app.
+
+## Plans
+
+A plan, roadmap or proposal can arrive as an interactive HTML page. **Push remote** hosts it at a shareable address, where you can annotate it. See [Shared page viewers](/help-center/apps/shared-page-viewers).
+
+## Media
+
+Generated images, video and audio show as preview cards with **Open** and **Download**. Open in [Canvas](/help-center/apps/canvas) to keep editing.
+
+## App cards
+
+- **Travel** results, with a detail panel.
+- **Shopping** products, with a detail panel and the cart.
+- **Events** from yantra.live.
+- **Workflow** cards that open the composer.
+- **Design**, **Study**, **Paper2Poster** and **Resume** results.
+
+## Code changes
+
+When the agent edits files, a diff card shows what changed.
+
+## Tool timeline
+
+A timeline lists every tool the agent used on a turn, so you can see what it read and did.
+
+## Related
+
+- [Chat Overview](/help-center/chat/chat-overview)

package/lib/templates/content/chat/chat-overview.md

@@ -0,0 +1,67 @@
+---
+title: 'Chat Overview'
+description: 'The Yantra chat: composer modes, sending and formatting, message actions, branching, temporary chats, history, and how apps open beside the conversation.'
+author: 'Yantra Team'
+---
+
+# Chat Overview
+
+Chat is the front door to Yantra. Ask anything, attach files, speak instead of type, pick a model, mention a connector, activate a skill, and watch apps open beside the conversation when they are needed.
+
+## Starting a chat
+
+Choose **Chat** in the sidebar or open `/new`. Every conversation has a stable address of the form `/c/<chat-id>`, and an app opened beside it adds the app name: `/c/<chat-id>/travel`.
+
+## The composer
+
+The placeholder reads "Ask anything". **Enter** sends, **Shift+Enter** adds a new line.
+
+Three modes sit above the composer:
+
+| Mode       | What it does                                                                    |
+| ---------- | ------------------------------------------------------------------------------- |
+| **Chat**   | A normal conversation with the agent and its tools                              |
+| **Search** | Deep Search: a multi-source research run with a progress timeline and citations |
+| **Build**  | Workflow mode, with the template grid for designs, decks, sites and media       |
+
+Around the composer you will find the model picker, attachments, the microphone, the **Channels** button, the skills picker, **Choose Template**, **Enhance prompt**, and a credits chip that shows what is left in your allowance.
+
+## Message actions
+
+Under each response: **Copy**, **Read aloud** (and Stop), **Try again** to regenerate, **Branch into new chat**, and **Share**. A branched chat shows a banner that says where it came from.
+
+## History and temporary chats
+
+The chat history sidebar lists your conversations, with a delete action per conversation.
+
+A **temporary chat** is never saved to history. It survives a page refresh, disappears when the tab closes, and keeps at most 200 messages.
+
+## Apps beside the chat
+
+When you ask for something an app handles, Yantra opens it in the right rail. "Find flights to Lisbon" opens Travel; "record my screen" opens Clips. You can also open any app from the launcher in the top bar. See [Apps Overview](/help-center/apps/apps-overview).
+
+When a turn is routed to your cloud Computer, a strip appears in the chat: **Working on your Computer** with a Watch link, then **Done on your Computer**. See [Computer](/help-center/apps/computer).
+
+## Approvals
+
+Before the agent runs a command it asks: **Allow**, **Allow all** for the rest of the session, or **Skip**. When the agent needs a decision from you it shows numbered options and a free-text answer. See [Approvals and agent questions](/help-center/chat/approvals-and-agent-questions).
+
+## Which agent runs the chat
+
+A gateway selector chooses the agent backend behind the conversation: the standard Yantra agent, or an [OpenClaw](/help-center/openclaw/openclaw-overview) gateway when your workspace has one. Some channels and features are available only on one of them.
+
+## Keyboard
+
+- Enter to send, Shift+Enter for a new line.
+- Esc closes dialogs.
+- In the @ mention picker: arrow keys to move, Enter to choose, Esc to dismiss.
+
+## Related
+
+- [Models and providers](/help-center/chat/models-and-providers)
+- [Attachments](/help-center/chat/attachments)
+- [Voice input and read aloud](/help-center/chat/voice-input-and-read-aloud)
+- [Sharing and exporting chats](/help-center/chat/sharing-and-exporting-chats)
+- [Connectors, skills and templates in chat](/help-center/chat/connectors-skills-and-templates)
+- [Deep Search](/help-center/chat/deep-search)
+- [Cards and widgets](/help-center/chat/cards-and-widgets)

package/lib/templates/content/chat/connectors-skills-and-templates.md

@@ -0,0 +1,34 @@
+---
+title: 'Connectors, Skills and Templates in Chat'
+description: 'Mention a connector with @, choose search sources, activate skills, pick a template, enhance a prompt, and connect an account in place when the agent needs it.'
+author: 'Yantra Team'
+---
+
+# Connectors, Skills and Templates in Chat
+
+The composer gives you four ways to shape what the agent uses on a turn.
+
+## @ mentions
+
+Type `@` to list connectors such as Gmail, Figma or GitHub. Pick one to bring it into the turn. A connector you have not connected yet can be connected right there, without leaving the chat. When the agent needs a site login, the connect card offers **Save to vault** so the login is kept in your project vault for next time.
+
+## Search sources
+
+The sources toggle chooses where Deep Search looks: web, GitHub, Reddit, YouTube, X and academic sources. The list comes from the search extensions installed in your workspace.
+
+## Skills
+
+The skills picker lets you select one or more skills and **Activate** them for the conversation. Many skills also activate on their own when you ask for what they do, or by their alias such as `/caveman`. See [Skills Overview](/help-center/skills/skills-overview).
+
+## Templates
+
+**Choose Template** opens the template gallery. A template seeds the composer with its prompt and the skills and connectors it needs. Remix or preview before you use it. See [Templates](/help-center/apps/templates).
+
+## Enhance prompt
+
+**Enhance prompt** rewrites what you typed into a fuller brief before sending.
+
+## Related
+
+- [Connectors overview](/help-center/connectors/connectors-overview)
+- [Chat Overview](/help-center/chat/chat-overview)

package/lib/templates/content/chat/deep-search.md

@@ -0,0 +1,30 @@
+---
+title: 'Deep Search'
+description: 'The Search composer mode: a multi-source research run with a progress timeline, citation pills, and sources you choose.'
+author: 'Yantra Team'
+---
+
+# Deep Search
+
+Switch the composer to **Search** to run Deep Search. Instead of one answer, you get a research run that searches, reads and synthesizes, with a live timeline and citations you can check.
+
+## What happens
+
+1. Yantra searches the web and any sources you toggled on.
+2. It crawls the pages that matter and reads them.
+3. It writes an answer with citation pills. Each pill opens the source.
+
+The progress timeline shows each step as it happens.
+
+## Sources
+
+Choose sources with the toggle beside the composer: web, GitHub, Reddit, YouTube (with transcripts), X and academic sources. Article content is pulled from URLs you paste.
+
+## Notifications
+
+Long research runs can finish while you are away. Turn on the email or web push notification for "when your research is ready" under [Notifications](/help-center/account-management/notifications).
+
+## Related
+
+- [Chat Overview](/help-center/chat/chat-overview)
+- [Connectors, skills and templates in chat](/help-center/chat/connectors-skills-and-templates)

package/lib/templates/content/chat/models-and-providers.md

@@ -0,0 +1,51 @@
+---
+title: 'Models and Providers'
+description: 'The model picker: Yantra Router tiers that need no key, models from Groq, Anthropic, OpenAI, Google, xAI and Moonshot, and bringing your own keys including Amazon Bedrock.'
+author: 'Yantra Team'
+---
+
+# Models and Providers
+
+The model picker sits beside the composer. It has three parts: **Select model**, **Saved keys** for keys you bring, and **Add provider**.
+
+## Default
+
+New chats use **Yantra Medium (Balanced)**. If that route is unavailable the chat falls back to Llama 3.3 70B on Groq.
+
+## Yantra Router
+
+The router tiers need no API key and are covered by your [credits](/help-center/plans-credits/ai-credits):
+
+| Tier          | Best for              |
+| ------------- | --------------------- |
+| **Small**     | Fast and efficient    |
+| **Medium**    | Balanced, the default |
+| **Complex**   | High quality          |
+| **Reasoning** | Deep thinking         |
+
+## Models by provider
+
+| Provider           | Models                                                                                                                                                 |
+| ------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------ |
+| **Groq**           | Llama 3.1 8B, Llama 3.3 70B, Llama 4 Scout 17B                                                                                                         |
+| **Anthropic**      | Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4, Claude Haiku 4.5                                                                                    |
+| **OpenAI**         | GPT-5.4, GPT-5.4 Pro, GPT-5.4 Mini, GPT-5.4 Nano                                                                                                       |
+| **Google**         | Gemini 3.1 Pro, Gemini 3 Flash                                                                                                                         |
+| **xAI**            | Grok 4.20, Grok 4.1 Fast, Grok 4.20 Reasoning, Grok Code Fast                                                                                          |
+| **Moonshot**       | Kimi K2.5, Kimi K2 Thinking, Kimi K2                                                                                                                   |
+| **Amazon Bedrock** | Claude Sonnet 4.6, Claude Opus 4.6, Nova Pro, Mistral Large 3, Llama 4 Maverick, MiniMax M2.5, DeepSeek R1, Nemotron 3 Super 120B. Bring your own key. |
+
+The list your workspace shows can differ, because providers and models are configured per workspace.
+
+## Bring your own key
+
+Open **Saved keys** to add a provider key, or **Add provider** to register a new one. Turns on your own key are billed by that provider, not from Yantra credits. Marketplace templates also accept a bring-your-own key during model configuration.
+
+## Media models
+
+Canvas and the media apps have their own model picker, including an **Auto** option that lets the media skill choose. Study's podcast feature picks text-to-speech providers separately.
+
+## Related
+
+- [AI credits](/help-center/plans-credits/ai-credits)
+- [Chat Overview](/help-center/chat/chat-overview)

package/lib/templates/content/chat/sharing-and-exporting-chats.md

@@ -0,0 +1,35 @@
+---
+title: 'Sharing and Exporting Chats'
+description: 'Share a single response or a whole conversation as a link, post it to X, LinkedIn or Reddit, and open the export in the viewer.'
+author: 'Yantra Team'
+---
+
+# Sharing and Exporting Chats
+
+Choose **Share** under any response to open the share dialog.
+
+## Scope
+
+- **Share response**: just this reply.
+- **Share conversation**: the whole thread.
+
+## Targets
+
+- **Copy link**
+- **X**
+- **LinkedIn**
+- **Reddit**
+
+The link points to a self-contained export page. **Open in viewer** shows it inside Yantra with the scope badge, share buttons and a feedback box. See [Shared page viewers](/help-center/apps/shared-page-viewers).
+
+## Unsharing
+
+A shared export can be unshared from the same dialog. The link stops working for everyone.
+
+## Branching
+
+To continue a conversation in a new direction without sharing, use **Branch into new chat** on a response. The new chat carries a banner saying where it branched from.
+
+## Related
+
+- [Chat Overview](/help-center/chat/chat-overview)

package/lib/templates/content/chat/voice-input-and-read-aloud.md

@@ -0,0 +1,29 @@
+---
+title: 'Voice Input and Read Aloud'
+description: 'Dictate a message with the microphone, hear any response read aloud, and use local speech-to-text and hotkeys on the desktop app.'
+author: 'Yantra Team'
+---
+
+# Voice Input and Read Aloud
+
+## Dictation
+
+Press the microphone beside the composer and speak. A live visualizer shows that you are being heard. Your speech is transcribed into the composer, where you can edit it before sending.
+
+## Read aloud
+
+Every response has a **Read aloud** action. Playback streams, so long answers start speaking right away. Press **Stop** to end it.
+
+## On the desktop app
+
+The [desktop app](/help-center/desktop-mobile/desktop-app) adds local speech-to-text that runs on your machine, and global hotkeys: `Cmd+Shift+M` for the microphone, `Cmd+Shift+V` to toggle audio, and `Cmd+Shift+R` to hold, record and send when you release. Use `Ctrl` instead of `Cmd` on Linux and Windows.
+
+## Voice in the apps
+
+- [Voicebox](/help-center/apps/voicebox) turns scripts into speech in many voices and transcribes recordings with Whisper.
+- [Study](/help-center/apps/study-notebook) turns a notebook into a two-speaker podcast.
+- The [browser side panel](/help-center/apps/sidecar-browser-panel) supports dictation too.
+
+## Related
+
+- [Chat Overview](/help-center/chat/chat-overview)

package/lib/templates/content/connectors/cdecli-command-line.md

@@ -0,0 +1,89 @@
+---
+title: 'The cdecli Command Line'
+description: 'The command line for Yantra: sign in, install and run connectors, manage credentials, skills, workflows, background tasks and extensions.'
+author: 'Yantra Team'
+---
+
+# The cdecli Command Line
+
+`cdecli` is the command line behind Yantra's agent. Everything the agent does with connectors, skills and workflows can also be done from a terminal.
+
+## Sign in
+
+```
+cdecli auth login
+cdecli auth status
+cdecli self-update
+```
+
+## Connectors
+
+```
+cdecli connector catalog
+cdecli connector list
+cdecli connector install <name>
+cdecli connector actions <name>
+cdecli connector action-schema -c <connector> -a <ACTION>
+cdecli connector exec -c <connector> -a <ACTION> --params '<json>'
+cdecli search "<what you want to do>"
+```
+
+`exec` accepts `--credential-id`, `--local`, `--runtime node` and `--timeout`.
+
+## Credentials
+
+See [Managing credentials](/help-center/connectors/managing-credentials).
+
+## Skills
+
+```
+cdecli skills list | search <query> | show <name>
+cdecli skills install <source> | install --local ./dir
+cdecli skills quick-publish | publish | remove <name>
+```
+
+## Workflows
+
+```
+cdecli workflow list | get <slug> | create | create-from-json | update-from-json
+cdecli workflow inputs <slug>
+cdecli workflow run <slug> --ver <version> [--wait] [--local] [--file <inputs>]
+cdecli workflow status | monitor | jobs | cleanup | delete
+```
+
+`wf` is an alias for `workflow`.
+
+## Background tasks
+
+```
+cdecli task run --label <label> -- <command>
+cdecli task list
+cdecli task logs <id> --follow
+cdecli task status <id>
+cdecli task cancel <id>
+```
+
+## Extensions and marketplace
+
+```
+cdecli extension list
+cdecli extension publish
+cdecli extension approve --slug <slug>
+cdecli extension public --slug <slug> --enable
+cdecli marketplace templates create -f <file>
+```
+
+## Remote agent
+
+Run commands on the hosted agent instead of locally:
+
+```
+cdecli agent exec -- connector list
+cdecli interact-agent -m "<prompt>"
+```
+
+## Related
+
+- [Connectors overview](/help-center/connectors/connectors-overview)
+- [Creating and publishing skills](/help-center/skills/creating-and-publishing-skills)
+- [Workflow Studio](/help-center/apps/workflow-studio)

package/lib/templates/content/connectors/connectors-overview.md

@@ -0,0 +1,40 @@
+---
+title: 'Connectors Overview'
+description: 'What a connector is, the catalog of more than a thousand apps, how to connect one from chat or settings, and how credentials, OAuth and API keys fit together.'
+author: 'Yantra Team'
+---
+
+# Connectors Overview
+
+A connector is Yantra's link to another app: Gmail, Slack, GitHub, HubSpot, Google Calendar, Stripe, Twilio, Notion and about a thousand more. Each connector exposes actions the agent can run and, for many, triggers that fire when something happens in that app.
+
+## The catalog
+
+The catalog holds roughly 1,100 connectors, grouped by category. Browse it on the Connectors page under your account settings, or ask the agent: "search connectors for invoicing". From the command line, `cdecli connector catalog` and `cdecli search "<what you want to do>"` do the same.
+
+## Connecting
+
+1. **From chat**: type `@` and pick the connector. If it is not connected, connect it right there.
+2. **From settings**: open Connectors, pick the connector, and complete OAuth or paste an API key.
+3. **From a connect card**: when the agent needs a connector mid-turn, it shows one.
+
+A connected account becomes a **credential**. See [Managing credentials](/help-center/connectors/managing-credentials).
+
+## Actions and triggers
+
+- **Actions** are things the agent does: send a message, create a record, list events.
+- **Triggers** deliver events from the app to your endpoint. See [Connector Triggers](/help-center/connectors/connector-triggers).
+- **Batch actions** run many operations in one call. See [Batch Actions](/help-center/connectors/batch-actions).
+
+## Yantra's own connectors
+
+Many apps are backed by Yantra connectors: shopping search, flight search, resortifi for hotels, resume builder, meetings, notebook, voicebox, videox, screen recorder, image, video and audio generation, storage, social ads and posts, mail campaign, inbox, card wallet, expenses, computer, clockbook, web design, webbuilder and workspaces. They connect the same way.
+
+## When a connector breaks
+
+Connector fixes ship as patches through a self-heal process, so a broken connector can be repaired without waiting for an upstream release. See [Self-heal](/help-center/connectors/self-heal).
+
+## Related
+
+- [Integrations Overview](/help-center/integrations/integrations-overview)
+- [Connectors and the vault](/help-center/account-management/connectors-and-vault)

package/lib/templates/content/connectors/managing-credentials.md

@@ -0,0 +1,42 @@
+---
+title: 'Managing Credentials'
+description: 'List, create, rotate and remove the credentials behind your connectors, in settings or from the command line.'
+author: 'Yantra Team'
+---
+
+# Managing Credentials
+
+A credential is a connected account: the OAuth grant or API key a connector uses on your behalf. Credentials are scoped to your project.
+
+## In settings
+
+Open Connectors under your account settings. Each connected connector lists its credentials. Reconnect to refresh a grant, or remove a credential you no longer want.
+
+## From the command line
+
+```
+cdecli credential ls
+cdecli credential list --provider <name>
+cdecli credential create --provider <name> --name "<label>"
+cdecli credential update --id <id> --rotate
+cdecli credential templates
+```
+
+Some connectors take a bearer token directly:
+
+```
+cdecli auth apikey -c <connector> --key "<token>" --auth-type bearer
+```
+
+## Multiple accounts
+
+A connector can have several credentials, for example two Gmail accounts. When the agent runs an action it uses the credential you name, or asks which one to use.
+
+## Site logins
+
+Website usernames and passwords are not credentials. They go in the project vault. See [Connectors and the vault](/help-center/account-management/connectors-and-vault).
+
+## Related
+
+- [Connectors overview](/help-center/connectors/connectors-overview)
+- [The cdecli command line](/help-center/connectors/cdecli-command-line)

package/lib/templates/content/connectors/marketplace-and-extensions.md

@@ -0,0 +1,44 @@
+---
+title: 'Marketplace and Extensions'
+description: 'The public extension gallery, the marketplace pages for extensions, publishers and templates, the AI generation templates, and the extensions that ship with Yantra.'
+author: 'Yantra Team'
+---
+
+# Marketplace and Extensions
+
+Extensions add connectors, templates, apps and skills to a workspace. The marketplace is where they are published, reviewed and installed.
+
+## Pages
+
+| Page              | Path                             | What it is                                       |
+| ----------------- | -------------------------------- | ------------------------------------------------ |
+| Gallery           | `/gallery`                       | The public extension gallery. No sign-in needed. |
+| Extensions        | `/marketplace/extensions`        | Browse and install                               |
+| Publishers        | `/marketplace/publishers`        | Who publishes what                               |
+| Publish extension | `/marketplace/publish-extension` | Submit your own                                  |
+| Form templates    | `/marketplace/form-templates`    | The template forms                               |
+| Review queue      | `/marketplace/review-queue`      | Pending submissions, for reviewers               |
+| Monetization      | `/marketplace/monetization`      | Paid extensions                                  |
+| Extension health  | `/marketplace/extension-health`  | Status of installed extensions                   |
+
+## AI generation templates
+
+The marketplace ships templates for audio, design, design systems, documents, Figma, images, live artifacts, mail, mobile apps, slide decks, travel, video and wireframes. Each is a four-step form: generation details, model configuration with an optional bring-your-own key, generation, and metadata. They appear in the [Templates](/help-center/apps/templates) gallery and the Build composer mode.
+
+## Extensions that ship with Yantra
+
+- **Search**: article extraction, Reddit, YouTube with transcripts, GitHub, and the web crawlers behind [Deep Search](/help-center/chat/deep-search).
+- **Media**: image generation and editing, video generation and editing, audio generation and editing, slides and carousels.
+- **Agents**: the standard agent gateway, a local agent for the desktop app, and the [OpenClaw](/help-center/openclaw/openclaw-overview) gateway.
+- **Characters**: Aria and Seraphina.
+- **Desktop**: local speech-to-text and local mouse and keyboard control.
+- **Chat**: a watcher that keeps in-progress replies alive when you switch conversations.
+
+## Apps as extensions
+
+An app can be published to a workspace as an extension, with no update to Yantra itself. See [Apps Overview](/help-center/apps/apps-overview).
+
+## Related
+
+- [Creating and publishing skills](/help-center/skills/creating-and-publishing-skills)
+- [Connectors overview](/help-center/connectors/connectors-overview)

package/lib/templates/content/connectors/self-heal.md

@@ -0,0 +1,30 @@
+---
+title: 'Self-heal for Connectors'
+description: 'How a broken connector gets fixed: self-heal files a patch, the fix is versioned and malware-scanned, and the repaired connector reaches everyone.'
+author: 'Yantra Team'
+---
+
+# Self-heal for Connectors
+
+Connectors are code, and code breaks when an upstream API changes. Self-heal is how a fix ships fast.
+
+## How it works
+
+1. When a locally installed connector fails, `cdecli self-heal` builds a fix as a patch, or a full source submission with `--source`, and files it as a pull request to the patches repository.
+2. Each submission is stamped with the connector version it was verified against. A submission for a stale version is closed automatically.
+3. Merged fixes are malware-scanned and forwarded to the connector source, which republishes the connector.
+4. The repaired connector reaches every user on the next install or update.
+
+## Applying a patch yourself
+
+Patches are unified diffs rooted at the connector directory under `~/.cdecli/connectors/<connector-id>`:
+
+```
+patch -p1 < patches/<connector-id>/<name>.patch
+patch -R -p1 < patches/<connector-id>/<name>.patch
+```
+
+## Related
+
+- [Connectors overview](/help-center/connectors/connectors-overview)
+- [The cdecli command line](/help-center/connectors/cdecli-command-line)