The Orkas Blog explains how Orkas works as an open-source, local-first desktop client for coordinating AI agents with your own model keys. These engineering notes cover the Agent Harness runtime, multi-agent orchestration, local self-evolution, provider routing, memory, context compaction, and the security choices behind direct-to-provider model traffic. Each article is written as a practical reference for developers, operators, and AI builders evaluating local-first AI agent workflows. Orkas 博客解释 Orkas 如何作为开源、本地优先的桌面客户端，用你自己的模型密钥协调 AI agent 团队。这里覆盖 Agent Harness 运行时、多 Agent 编排、本地自演进、Provider 路由、记忆、上下文压缩，以及模型流量直连供应商的安全设计。 Orkas Blog は、Orkas が自分のモデルキーで AI エージェントチームを調整する、オープンソースでローカルファーストなデスクトップクライアントとしてどう動くかを解説します。 O Blog do Orkas explica como o Orkas funciona como um cliente desktop open source e local-first para coordenar agentes de IA com suas próprias chaves de modelo. Estas notas de engenharia cobrem o runtime Agent Harness, orquestração multiagente, autoevolução local, roteamento de provedores, memória, compactação de contexto e as decisões de segurança por trás do tráfego direto ao provedor.
How the desktop runtime turns model calls into reliable agent sessions: streaming loops, tool routing, crash-safe state, provider abstraction, and context management.桌面运行时如何把模型调用变成可靠的 agent 会话：流式循环、工具路由、状态恢复、Provider 抽象和上下文管理。モデル呼び出しを信頼できるエージェントセッションに変える実装。Como o runtime desktop transforma chamadas de modelo em sessões confiáveis de agentes: loops em streaming, roteamento de ferramentas, estado tolerante a falhas, abstração de provedores e gestão de contexto.
How a lead agent decomposes a request, dispatches sub-agents, passes context between steps, and recovers when a task fails.主 agent 如何拆解请求、派发子 agent、在步骤间传递上下文，并在失败后恢复。リードエージェントがタスクを分解し、サブエージェントを動かす仕組み。Como um agente líder decompõe uma solicitação, despacha subagentes, passa contexto entre etapas e se recupera quando uma tarefa falha.
Why Orkas keeps workspace data and model keys on the user’s machine, and why traffic from your own provider goes directly to it rather than through Orkas servers.为什么 Orkas 把工作区数据和模型密钥留在用户机器上，以及为什么使用自己的供应商时模型流量会直连该供应商、不经 Orkas 服务器代理。データとキーを手元に置き、自分のプロバイダーを使う場合に通信が Orkas のサーバーを経由せず直接送られる理由。Por que o Orkas mantém dados do workspace e chaves de modelo na máquina do usuário e por que o tráfego do seu próprio provedor vai diretamente a ele, sem passar pelos servidores do Orkas.
Getting cited is not ranking. It's surviving retrieval, then being the passage worth quoting — the three OpenAI bots, the CDN gate robots.txt hides, and why a fact behind JavaScript doesn't exist.
Claude Code and Codex each win at different things. Run them together — with terminals and git worktrees, or from one Orkas Commander that orchestrates both in a single chat.
How Orkas syncs user data across devices with encrypted transfer, content storage, server-owned commits, account locks, sync rules, model-assisted conflict handling, delete confirmation, and a recycle bin.
How Orkas rebuilt its agent foundation across the 1.0 release line: an in-process runtime, provider rotation, dynamic group-chat orchestration, open hosting, memory, and self-evolution.
Inside Orkas's multi-agent orchestration: a lead agent turns one request into a plan, dispatches sub-agents by dependency, passes context between steps, and heals from failure.
What local-first AI means — your data, API keys, and model traffic stay on your own machine, not a vendor's cloud — why it matters for privacy, and how a bring-your-own-key agent actually works.
How Orkas turns model calls into a reliable desktop agent runtime: streaming run loops, tool routing, context compaction, provider abstraction, memory, and crash-safe sessions.
Inside Orkas's local self-evolution loop: lightweight signals, background reflection, executable skills, skill metrics, and guardrails against learning the wrong lesson.
Orkas is an open-source, local-first desktop AI client for macOS and Windows. Instead of chatting with a single assistant, you direct a team of agents: a lead agent owns your goal and recruits sub-agents that call skills to do the work. You can use optional Orkas-managed official models or connect your own provider through OAuth or an API key; your workspace is local-first by default.Orkas é um cliente de IA desktop open source, local-first para macOS e Windows. Em vez de conversar com um único assistente, você dirige uma equipe de agentes: um agente líder é dono do seu objetivo e recruta subagentes que convocam habilidades para fazer o trabalho. Você pode usar modelos oficiais opcionais gerenciados pelo Orkas ou conectar seu próprio provedor por OAuth ou chave de API; seu espaço de trabalho é local-first por padrão.Orkas 是一个开源、本地优先的桌面 AI 客户端，支持 macOS 和 Windows。你不是和单个助手对话，而是指挥一支 agent 团队：主 agent 负责你的目标，招募子 agent 调用技能来完成工作。你可以选择使用 Orkas 托管的官方模型，也可通过 OAuth 或 API Key 接入自己的供应商；工作区默认本地优先。Orkas は macOS と Windows 向けの、オープンソースでローカルファーストなデスクトップ AI クライアントです。単一のアシスタントと話す代わりに、エージェントのチームを指揮します。リードエージェントが目標を担い、サブエージェントを招集してスキルを呼び出し作業を進めます。任意の Orkas 管理公式モデルを使うか、OAuth または API キーで自分のプロバイダーを接続できます。ワークスペースは既定でローカルファーストです。
Local-first means your data and control stay on your device. Your chats, files, knowledge base, memory, and encrypted model keys live on your machine by default. With your own provider, model traffic goes directly from your computer to that provider and is not proxied through Orkas servers; official models use Orkas's managed model service.Local-first significa que seus dados e controle permanecem no seu dispositivo. Seus chats, arquivos, base de conhecimento, memória e chaves de modelo criptografadas ficam em sua máquina por padrão. Com seu próprio provedor, o tráfego de modelo vai diretamente do seu computador para ele e não passa pelos servidores do Orkas como proxy; os modelos oficiais usam o serviço de modelo gerenciado do Orkas.本地优先意味着数据和控制权都留在你的设备上。默认情况下，你的对话、文件、知识库、记忆和加密后的模型密钥都存在本机。使用自己的供应商时，模型流量从你的电脑直连该供应商，不经 Orkas 服务器代理；官方模型使用 Orkas 的托管模型服务。ローカルファーストとは、データと制御が自分の端末に残るということです。チャット、ファイル、ナレッジベース、メモリ、暗号化されたモデルキーは既定で手元に保存されます。自分のプロバイダーを使う場合、モデル通信はあなたのPCからそのプロバイダーへ直接送られ、Orkas のサーバーを経由しません。公式モデルは Orkas のマネージドモデルサービスを利用します。
A lead agent reads your goal, breaks it into steps, and recruits sub-agents by task and capability. Each sub-agent works in its own bounded context and calls skills — web search, code execution, file I/O, knowledge-base search, and connectors — to deliver. The lead passes each sub-agent only what it needs, which keeps token cost down and responsibilities clean.Um agente líder lê sua meta, divide-a em etapas e recruta subagentes por tarefa e capacidade. Cada subagente trabalha em seu próprio contexto limitado e chama habilidades – pesquisa na web, execução de código, E/S de arquivos, pesquisa na base de conhecimento e conectores – para entregar. O lead passa a cada subagente apenas o que ele precisa, o que mantém o custo do token baixo e as responsabilidades limpas.主 agent 读取你的目标，拆成步骤，并按任务和能力招募子 agent。每个子 agent 在各自受限的上下文中工作，调用技能——网页搜索、代码执行、文件读写、知识库检索和连接器——来交付结果。主 agent 只把必要的信息传给每个子 agent，从而降低 token 成本、保持职责清晰。リードエージェントが目標を読み取り、ステップに分解し、タスクと能力に応じてサブエージェントを招集します。各サブエージェントは限定されたコンテキストで働き、スキル（Web 検索・コード実行・ファイル入出力・ナレッジベース検索・コネクタ）を呼び出して成果を出します。リードは必要な情報だけを渡すため、トークンコストを抑え、責務を明確に保てます。
Common uses include research and analysis, writing and editing, coding with native or external CLI agents such as Claude Code, Codex, and OpenClaw, data work, learning, and office documents. You can also turn a recurring task into a reusable sub-agent once and summon it in chat whenever you need it.Os usos comuns incluem pesquisa e análise, escrita e edição, codificação com agentes CLI nativos ou externos, como Claude Code, Codex e OpenClaw, trabalho de dados, aprendizagem e documentos de escritório. Você também pode transformar uma tarefa recorrente em um subagente reutilizável uma vez e invocá-lo no chat sempre que precisar.常见用途包括研究与分析、写作与编辑、用原生或外部 CLI agent（如 Claude Code、Codex、OpenClaw）编程、数据处理、学习，以及办公文档。你也可以把一个重复任务一次性做成可复用的子 agent，之后在对话里随时召唤。主な用途は、調査と分析、執筆と編集、ネイティブまたは外部 CLI エージェント（Claude Code、Codex、OpenClaw など）でのコーディング、データ作業、学習、オフィス文書などです。繰り返す作業を一度だけ再利用可能なサブエージェントにして、必要なときにチャットから呼び出すこともできます。
The current public release is a macOS and Windows desktop app; the iOS remote-control relay is disabled, and there is no web client. You can use optional Orkas-managed official models or connect a provider through OAuth or an API key. Your provider bills usage from your own key; managed capabilities use Orkas credits. Output quality depends on the model you connect. Optional multi-device sync stores synced data on Orkas servers, and the free edition sends limited usage analytics.A versão pública atual é um app desktop para macOS e Windows; o relay de controle remoto iOS está desabilitado e não há cliente web. Modelos oficiais usam créditos Orkas, ou você conecta seu provedor e é cobrado por ele. A sincronização opcional armazena os dados escolhidos no Orkas.当前公开版本是 macOS/Windows 桌面应用；iOS 远程控制中继已关闭，且没有网页客户端。可使用 Orkas 托管的官方模型或连接自有供应商；前者使用 Orkas 积分，后者由供应商计费。可选同步会把所选数据存到 Orkas。現在の公開版は macOS/Windows デスクトップアプリです。iOS リモートコントロールリレーは無効で、Web クライアントはありません。公式マネージドモデルまたは自分のプロバイダーを利用でき、任意同期は選択データを Orkas に保存します。
