问一句能回，真要交付就开始漏需求、漏步骤、漏验收，最后还是你来补和重做。
Your AI workforce. Scale your business.
Your AI team
Direct your AI team by chat, and scale your business.
A super-powered Commander leads a team of specialist agents that do the work for you — your AI workforce, ready whenever you need it.
Starting your download正在开始下载ダウンロードを開始していますIniciando o download Didn’t start, or need another platform? Open the download page.没有自动下载，或需要其他平台？请前往下载页面。開始しない、または別のプラットフォームが必要ですか？ダウンロードページをご覧ください。Não começou ou precisa de outra plataforma? Acesse a página de download. Didn’t start or was blocked? Try another browser or adjust system security; for other platforms, open the download page.没有自动下载或被拦截？请更换浏览器或调整安全设置；其他平台请前往下载页面。自動で開始しない、またはブロックされましたか？別のブラウザかセキュリティ設定をお試しください。別の版はダウンロードページへ。Não começou ou foi bloqueado? Tente outro navegador ou ajuste a segurança; para outras plataformas, acesse a página de download.
Download the app and start. Official models work out of the box, or connect your own.下载应用即可开始。官方模型开箱即用，也可接入你自己的模型。アプリをダウンロードしてすぐ開始。公式モデルをそのまま使うことも、自分のモデルを接続することもできます。Baixe o app e comece. Use os modelos oficiais imediatamente ou conecte os seus.
Commander + specialist agents Local-first by default MIT open source
Professional team
Team members
Start with built-in specialists, create members around your own work, or expand the team with marketplace and external agents.
Lead agent
Turn one goal into one usable delivery
Commander clarifies the outcome, chooses the right agents, skills, connectors, and tools, then keeps the work moving until the results come back together.
- Turns a vague request into clear goals, constraints, inputs, and success criteria.
- Runs independent work in parallel and dependent work in sequence.
- Reads every result, reroutes around gaps, and delivers one coherent answer or file.
VideoStudio
Create, edit, and localize video in one workflow
Start from a brief, reference image or video, or existing footage. VideoStudio creates motion graphics, analyzes and edits footage, adds captions and voice tracks, and combines supplied, generated, and composed segments into a reviewable video.
- Uses reference images and videos to guide, constrain, or edit the result.
- Routes work through compose, generate, edit, or a mixed production.
- Shows reviewable plans and previews, then reuses unaffected segments when you revise one part.
Video production4 routes · reviewable workflow
COMPOSE
Motion graphics
Designed explainers and animated UI
AUTO
Mixed production
Reference, generated, and edited segments
GENERATE
Footage & avatars
Original shots, characters, and b-roll
EDIT
Smart edit & localization
Trim, captions, dubbing, mixing, and highlights
plan.jsonpreview and confirm
00-08Title and hookcompose
08-34Product walkthroughfootage
34-50Generated b-rollgenerate
50-60CTA cardcompose
ImageStudio
Design, generate, and edit still visuals
ImageStudio makes posters, covers, social graphics, infographics, charts, illustrations, and precise edits—using HTML/CSS/SVG first and image models only when the look needs them.
- Chooses compose, hybrid, generate, or edit from the visual intent.
- Keeps copy and layout exact while using generated pixels only where useful.
- Protects reference details and runs a scored design review before export.
PptMaker
Turn source material into an editable, reviewable deck
Give it a topic, outline, documents, data, reference images, or an existing presentation. It plans the story and visual system, builds or safely edits native PowerPoint slides, and validates the rendered result.
- Follows authorized visual references, or derives art direction from the audience and communication goal.
- Keeps text, shapes, charts, and tables editable instead of flattening the deck into images.
- Checks every slide for content, design, and coherence, then repairs concrete rendering defects.
UIDesigner
Turn product context into an editable UI artifact
Give it a goal, PRD, current screen, screenshot, inspectable Figma material, or brand reference. It returns an HTML-first design you can review, revise, package, or hand to engineering.
- Preserves the real workflow and source structure instead of forcing a template.
- Adds only the responsive, accessibility, component, and runtime states the product needs.
- Can produce live dashboards, design systems, and formal UI reviews when needed.
OfficeWorker
Create, edit, and deliver every office format
It creates, reads, edits, checks, and delivers Word, Excel, PowerPoint, WPS-compatible files, and PDFs as one coordinated office workflow.
- Works from source materials or existing Word, Excel, PowerPoint, WPS, and PDF files.
- Preserves structure, formulas, layouts, comments, and source files.
- Checks formatting, compatibility, page order, and export before handoff.
ContentWriter
From source pack to publish-ready long-form
For articles, blogs, newsletters, tutorials, cases, and opinion pieces, it builds a brief and source/claim ledger, drafts and restructures, calibrates the voice, removes AI stiffness, and audits facts.
- Matches the angle, audience, format, and source boundary before drafting.
- Revises structure and voice instead of merely polishing sentences.
- Audits claims, citations, fact risk, and missing sources before delivery.
ProductDeveloper
Ship features, fixes, and refactors with evidence
Starting from a PRD, requirement, issue, bug report, or design, ProductDeveloper maps the repository, makes focused changes, and proves each accepted behavior with fresh evidence.
- Handles features, bugs, CI/test repair, refactors, code review, and performance debugging.
- Keeps patches focused while checking architecture, compatibility, migration, and rollback risk.
- Maps every acceptance point to tests, review evidence, or an explicit unverified item.
SeoGeoAgent
Audit, fix, and monitor SEO/GEO with evidence
Give it a URL for technical SEO, content quality, schema, Core Web Vitals, and GEO citation-readiness checks, with a health score, dashboard, and prioritized action plan.
- Runs deep diagnosis, quick checks, monitoring, source fixes, or answer-first content work.
- Uses connected Search Console or Bing Webmaster data to separate real signals from estimates.
- Can edit a provided local repository, retest affected pages, and report verified changes.
DeepResearcher
Evidence-grounded research with confidence and limits
DeepResearcher plans the question, combines your materials with web and scholarly sources, compresses the evidence, checks citations, and delivers an auditable report.
- Bounds the scope and sub-questions before collecting sources.
- Separates verified evidence, sampled signals, inference, contradictions, and gaps.
- Reports source quality, confidence, limitations, and references you can verify.
Create your own team member
Start with a name and a one-line responsibility. Orkas helps refine how the member works and what it needs—or turns a useful conversation directly into a reusable agent.
1Name & role 2Refine capabilities 3Call anytime
Get Orkas and create a member →More agents in the marketplace
Discover more specialists for ecommerce, product and research, creation, office work, growth, learning, data, and other workflows.
Ecommerce Product & Research Creation Office Growth Learning Data More
Open marketplace →Plug in external agents
Connect Claude Code, Codex, OpenClaw, OpenCode, Hermes Agent, and more as external coding agents that work alongside the built-in team.
CCClaude Code CXCodex CLOpenClaw OCOpenCode HAHermes Agent +More
Use cases
What you can do with Orkas
Start from a real work goal. Commander combines specialist agents, your local workspace, and optional platform capabilities into one delivery.
01
Market entry decision
A product team must decide which market segment to enter, using public sources, customer notes, and a reviewable evidence trail.
"Research this market, compare the leading products, and deliver a cited recommendation for our review."
DeepResearcher→ContentWriter→OfficeWorker
View market-entry scenario → 02
Product launch campaign
A launch owner needs the article, social visuals, and product video to tell the same story from one approved source pack.
"Create the launch article, social visuals, and a 60-second product video from this brief."
ContentWriter→ImageStudio→VideoStudio
View launch-campaign scenario → 03
Product workflow launch
An operations requirement must become an editable interface, tested code, and a product page that search and AI engines can understand.
"Build this operations dashboard, cover loading and error states, run the tests, and optimize the product pages for SEO/GEO."
UIDesigner→ProductDeveloper→SeoGeoAgent
View product-workflow scenario → 04
Internal knowledge desk
A support or operations team needs fast answers from approved policies, reports, notes, and manuals—with page-level citations.
"Organize these project files, answer from the original sources, and prepare an executive brief with page citations."
DeepResearcher→OfficeWorker→PptMaker
View knowledge-desk scenario → 05
Monthly business review
Finance and operations need one reconciled story across meeting notes, the sales workbook, the Word report, the review deck, and the final PDF.
"Use these meeting notes, this sales workbook, and our template to update the analysis, write the Word report, build the review deck, and export the final PDF."
OfficeWorker→PptMaker→ContentWriter
View monthly-review scenario → 06
Weekly team operations
A team lead wants Monday updates collected from connected tools, project notes refreshed, and a status draft ready for review.
"Every Monday, summarize new issues and messages, update the project notes, and draft the team status for review."
Commander→OfficeWorker→ContentWriter
View weekly-operations scenario →Core strengths
Orkas highlights
用户常见痛点
调研、写方案、改代码、做视频分散在不同对话和工具里，谁做完了、下一步是什么全靠你记。
文件、聊天、知识库和生成结果存在哪里、谁能看、是否同步都说不清，真实资料只能继续留在外面。
偏好、项目规则、历史决定留不住，换个任务或隔天再开，还是要重新把上下文喂一遍。
只能用平台给的模型和会员额度，已有 API Key、公司供应商和成本控制都接不上。
CLI、MCP、Agent 各自能跑，但结果散在各处，下一次同类任务还是从零开始。
Orkas 的回答
所以 Orkas 把核心能力放在三件事上：可靠地组织交付、可控地处理数据和权限、开放地接入你的工具生态。
01
专业可靠
每次交付都从明确分工开始：先拆解目标，再协同推进，最后把经验沉淀下来。
02
Secure control
Chats, files, and knowledge materials stay on your device by default; sensitive settings are encrypted locally, and you can limit agent access and confirm critical actions.
Local storage
Chats, files, project history, and knowledge material stay on your device by default.
- 对话与文件
- 项目历史
- 本地资料库
Sensitive data encryption
API keys and sensitive settings are protected locally instead of being exposed as plain text.
- Key 加密保存
- 本机凭据管理
- 降低明文暴露
Access permissions
You decide what an agent can read, call, or change, so different work can have different boundaries.
- 可读资料范围
- 可调用工具
- 可修改边界
Critical action confirmation
Before risky operations, Orkas can pause and ask you to confirm the action and cost.
- 高风险动作
- 成本消耗提示
- 人工确认
03
Open extension
Orkas works with your models, external agents, and existing systems instead of locking you into one platform.
Open source code
You can inspect how it works, self-host what you need, and avoid being locked into a black box.
- 代码可审计
- 可自托管
- 不做黑盒
Multiple model choices
支持主流模型和 Orkas 官方模型，统一配置后适配不同团队、成本和能力需求。
- Orkas 官方模型
- 自带主流模型 Key
- 供应商可切换
External agent access
External coding agents and command-line tools can be called from the same conversation when needed.
- 编码 Agent
- 本地 CLI 工具
- 同一对话调用
MCP and open ecosystem
Connect data sources, tools, and community-built capabilities through MCP and compatible integrations.
- 数据源接入
- 工具连接器
- 社区能力
FAQ
Frequently asked questions
What is Orkas?Orkas 是什么？Orkas とは何ですか？O que é Orkas?
Orkas is an open-source, local-first AI workforce. A super-powered Commander coordinates specialist agents to complete complex work together.Orkas 是开源、本地优先的 AI 生产力团队。超强指挥官协调专家 Agent，共同完成复杂工作。Orkas はオープンソースでローカルファーストな AI ワークフォースです。強力な Commander が専門エージェントを調整し、複雑な仕事を協力して完了させます。Orkas é uma força de trabalho de IA open source e local-first. Um Commander superpoderoso coordena agentes especialistas para concluir trabalhos complexos juntos.
What can I build with Orkas?用 Orkas 能做什么？Orkas で何ができますか？O que posso fazer com Orkas?
Use Orkas as a coordinated AI team for complex work: research markets, turn findings into reports or slides, read PDFs with citations, fix bugs and open PRs, analyze data, create apps or videos, organize inbox and calendar work, connect external tools, and more, then turn repeated work into reusable agents and skills.你可以把 Orkas 当成一支可调度的 AI 团队：做市场和竞品调研，生成报告或演示稿，阅读 PDF 并带引用回答，复现 bug 并提交修复 PR，分析表格和数据，生成交互式应用或视频，整理收件箱和日历，连接外部工具等；还可以把重复工作沉淀成可复用的 Agent 和技能。Orkas は複雑な仕事を進める AI チームとして使えます。市場や競合の調査、レポートやスライド作成、PDF の引用付き回答、バグ再現と修正 PR、データ分析、アプリや動画作成、受信箱やカレンダー整理、外部ツール連携などに加え、反復作業を再利用可能な Agent やスキルとして残せます。Use o Orkas como uma equipe de IA coordenada para trabalhos complexos: pesquisar mercados, transformar descobertas em relatórios ou slides, ler PDFs com citações, corrigir bugs e abrir PRs, analisar dados, criar apps ou vídeos, organizar inbox e calendário, conectar ferramentas externas e muito mais, além de transformar trabalho repetido em agentes e skills reutilizáveis.
How is Orkas different from ChatGPT, Codex, Claude Code, or Cline?Orkas 和 ChatGPT、Codex、Claude Code、Cline 有什么不同？Orkas は ChatGPT、Codex、Claude Code、Cline と何が違いますか？Como Orkas difere de ChatGPT, Codex, Claude Code ou Cline?
ChatGPT is a hosted assistant, while Codex, Claude Code, and Cline focus on coding sessions. Orkas is an open-source, local-first AI workforce where a Commander coordinates specialist agents to complete complex work together.ChatGPT 更像托管助手，Codex、Claude Code 和 Cline 主要聚焦编码会话。Orkas 是开源、本地优先的 AI 生产力团队，由指挥官协调专家 Agent 共同完成复杂工作。ChatGPT はホスト型アシスタントで、Codex、Claude Code、Cline は主にコーディングセッションに焦点があります。Orkas はオープンソースでローカルファーストな AI ワークフォースで、Commander が専門エージェントを調整し、複雑な仕事を協力して完了させます。ChatGPT é um assistente hospedado, enquanto Codex, Claude Code e Cline focam sessões de código. Orkas é uma força de trabalho de IA open source e local-first, onde um Commander coordena agentes especialistas para concluir trabalhos complexos juntos.
Do I need to bring my own model key?必须自带模型 Key 吗？自分のモデルキーが必要ですか？Preciso trazer minha própria chave de modelo?
No. Orkas can provide managed official models that run through Orkas-managed model service. You can also connect your own providers through OAuth or API keys, including OpenAI, Claude, Gemini, DeepSeek, Kimi, and GLM; bring-your-own-provider calls go directly to that provider.不必须。Orkas 可以提供通过 Orkas 托管模型服务运行的官方模型。你也可以通过 OAuth 或 API Key 连接自己的供应商，包括 OpenAI、Claude、Gemini、DeepSeek、Kimi 和 GLM；自带供应商的请求会直达该供应商。必須ではありません。Orkas 管理の公式モデルを使うこともできます。OpenAI、Claude、Gemini、DeepSeek、Kimi、GLM など自分のプロバイダーを OAuth または API キーで接続することもでき、その場合の呼び出しは直接そのプロバイダーへ送られます。Não. Orkas pode fornecer modelos oficiais gerenciados que passam pelo serviço de modelos gerenciado pelo Orkas. Você também pode conectar seus provedores por OAuth ou API key, incluindo OpenAI, Claude, Gemini, DeepSeek, Kimi e GLM; chamadas para provedores próprios vão direto a esse provedor.
Will my files, history, or API keys be uploaded to Orkas?我的文件、历史或 API Key 会上传到 Orkas 吗？ファイル、履歴、API キーは Orkas にアップロードされますか？Meus arquivos, histórico ou API keys serão enviados ao Orkas?
By default, your configs, history, knowledge bases, and generated files stay on your machine. If you bring your own keys, they are stored locally and requests go directly to your provider. Orkas-managed models and optional cloud features use Orkas services only for those selected features.默认情况下，配置、历史、知识库和生成文件都留在你的机器上。如果你自带 Key，Key 会本地保存，请求直达供应商。只有你选择官方托管模型或可选云功能时，对应功能才使用 Orkas 服务。標準では、設定、履歴、ナレッジベース、生成ファイルは自分のマシンに残ります。持ち込みキーはローカルに保存され、リクエストは直接プロバイダーへ送られます。Orkas 管理モデルや任意のクラウド機能は、選択した機能に限って Orkas のサービスを使います。Por padrão, configurações, histórico, bases de conhecimento e arquivos gerados ficam na sua máquina. Se você trouxer suas chaves, elas ficam locais e as chamadas vão direto ao provedor. Modelos gerenciados pelo Orkas e recursos opcionais de nuvem usam serviços Orkas apenas para esses recursos escolhidos.
Is Orkas free, and what would I pay for?Orkas 免费吗？我需要为什么付费？Orkas は無料ですか？何に支払いますか？Orkas é gratuito? Pelo que eu pagaria?
The open-source desktop edition is free to download and self-host. Bring-your-own-provider usage is billed by your own provider. Paid Orkas plans add hosted services such as managed models, sync, and cloud storage. Remote control and team collaboration are planned.开源桌面版可以免费下载和自托管。自带供应商的模型调用由你的供应商计费。Orkas 付费计划主要增加官方托管模型、同步和云存储等托管服务。远程控制和团队协作仍在规划中。オープンソースのデスクトップ版は無料でダウンロード、セルフホストできます。持ち込みプロバイダーの利用料は自分のプロバイダーに支払います。有料 Orkas プランでは、管理モデル、同期、クラウドストレージなどのホスト型サービスが追加されます。リモート操作とチーム共同作業は現在計画中です。A edição desktop open source é gratuita para baixar e auto-hospedar. Uso de provedores próprios é cobrado pelo seu provedor. Planos pagos do Orkas adicionam serviços hospedados como modelos gerenciados, sync e armazenamento em nuvem. Controle remoto e colaboração em equipe estão planejados.
Full comparison
Orkas vs others
Orkas is stronger when work needs a team, local context, reusable workflows, and control over model keys.
| Tool | Main use | Setup cost | Workspace scope | Coordination / reuse | Data / key control | Best fit |
|---|---|---|---|---|---|---|
| OrkasAI team desktop | Command an AI team across code, docs, research, data, and operations. | Desktop app, ready to use | Full workflow workspace | Commander + reusable sub-agents | Official models + BYO keys | You want complex work to become a reusable team workflow. |
| CodexCoding agent | Work on coding tasks inside a repository. | Client / CLI | Repository | Single coding agent session | Client/account based | You mainly need focused repository coding. |
| Claude CodeCoding agent | Work on deep coding tasks inside a repository. | Client / CLI | Repository | Single coding agent session | Client/account based | You mainly need focused repository coding. |
| ClineEditor coding agent | Autonomous coding inside VS Code. | VS Code plugin | Editor workspace | Single editor conversation | Configured in plugin | You want hands-on coding where you edit. |
| LangChainAgent framework | Build custom agent systems in code. | Engineering required | Whatever you build | You design orchestration | You implement control | You are building agent infrastructure. |
| OpenClawSelf-hosted assistant gateway | Run a personal assistant across channels, agents, and skills. | Self-hosted setup | Gateway + channels + workspaces | Agent routing + sub-agent runs | Self-operated | You want a channel-first, self-hosted personal assistant. |
| Hermes AgentAlways-on server agent | Run background and scheduled automations. | Server setup | Server workflows | Cron / scripts | Self-operated | You need unattended background automation. |
| DeepSeek HarnessComposable agent runtime | Build and run a plugin-composed agent for workspace tasks. | Node.js / npm developer preview | Workspace / custom runtime | Replaceable agent loop + subagents | Local config + provider keys | You want to compose or extend the agent runtime itself. |
Your AI team is ready
Install the desktop app, choose a managed official model or connect your own provider, and put an AI team to work in minutes.
Commander + specialist agents Local-first by default MIT open source