Chi Wang

Claude Cowork proves the interest in AI coworkers. Orion (meetorion.app) takes the autonomous working experience to another level: • Works with and enhances your experience with your favorite apps (messages, docs, emails, sheets, files, calendars…) and AI tools (ChatGPT, Claude Code, Gemini, Grok…) • Supervises workflows when you are away, from vibe coding to media creation • Coordinates AI agents across all your devices (macOS, windows, VM), each operating one device independently Your personal AI superpower. New surprising ways of vibe working every day - I just tried using Orion to add subtitles to the video and it worked! Watch it handle multiple tasks autonomously from a single cmd+k.

We're excited to release MassGen v0.1.63 — Round Evaluator Contracts! 🚀 MassGen now supports improved critique and toggling the level of changes in the round evaluator, with transformation pressure and success contracts for deeper quality. Plus: subagents run as an ensemble by default with lighter refinement, and killed agent handling. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.4, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Checklist-gated quality rounds with dynamic evaluation criteria drive iterative refinement. Spawn subagents for focused background work, including specialized roles like round evaluator with ensemble pattern defaults for delegated quality assessment. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution — Structured plans with chunked execution, progress checkpoints, iterative review, and verification. Spec mode for formal requirements before execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. Use MassGen as a skill from other AI coding agents. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Multiple providers (OpenAI, Google, ElevenLabs, Nano Banana 2, Grok Imagine) with multi-turn image editing and reusable media generation skills. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI & Web UI — Watch agents in real-time via Terminal UI with mode cycling, log analysis, tool batching, and keyboard shortcuts. Session viewer for observing automation runs. Also available: browser-based Web UI. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.62 — MassGen Skill & Viewer! 🚀 New general-purpose MassGen Skill (github.com/massgen/skills) with 4 modes (general, evaluate, plan, spec) for use from Claude Code and other AI agents. Session viewer for real-time observation of automation runs. Backend improvements and quickstart enhancements. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.4, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Checklist-gated quality rounds with dynamic evaluation criteria drive iterative refinement. Spawn subagents for focused background work, including specialized roles like round evaluator for delegated quality assessment. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution — Structured plans with chunked execution, progress checkpoints, iterative review, and verification. Spec mode for formal requirements before execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. Use MassGen as a skill from other AI coding agents. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Multiple providers (OpenAI, Google, ElevenLabs, Nano Banana 2, Grok Imagine) with multi-turn image editing and reusable media generation skills. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI & Web UI — Watch agents in real-time via Terminal UI with mode cycling, log analysis, tool batching, and keyboard shortcuts. Session viewer for observing automation runs. Also available: browser-based Web UI. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.61 — Round Evaluator Paradigm! 🚀 New specialized subagent type that automatically spawns evaluator subagents after each new answer to provide detailed feedback as input to the next round. Major orchestrator refactoring with improved evaluation prompts, task plan injection, and subagent fixes. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.4, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Checklist-gated quality rounds with dynamic evaluation criteria drive iterative refinement. Spawn subagents for focused background work, including specialized roles like round evaluator for delegated quality assessment. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution — Structured plans with chunked execution, progress checkpoints, iterative review, and verification. Spec mode for formal requirements before execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Multiple providers (OpenAI, Google, ElevenLabs, Nano Banana 2, Grok Imagine) with multi-turn image editing and reusable media generation skills. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI & Web UI — Watch agents in real-time via Terminal UI with mode cycling, log analysis, tool batching, and keyboard shortcuts. Also available: browser-based Web UI. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.59 — Quality Round Improvements! 🚀 Smarter planning with auto-added improvements and plan review. Agents now save replayable verification steps to a dedicated markdown file, auto-injected into future rounds so the next agent can replay the exact verification pipeline. Plus: checklist evaluation fixes, subagent enhancements, and media generation improvements. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Checklist-gated quality rounds with dynamic evaluation criteria drive iterative refinement. Spawn subagents for focused background work, including specialized roles. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution — Structured plans with chunked execution, progress checkpoints, iterative review, and verification. Spec mode for formal requirements before execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Multiple providers (OpenAI, Google, ElevenLabs, Nano Banana 2, Grok Imagine) with multi-turn image editing and reusable media generation skills. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI & Web UI — Watch agents in real-time via Terminal UI with mode cycling, log analysis, tool batching, and keyboard shortcuts. Also available: browser-based Web UI. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.58 — a Comprehensive Multimodal Revamp! 🚀 Three new media providers for comprehensive set (OpenAI GPT Image/Sora/audio, Google Nano Banana 2/Veo 3.1, Grok Imagine, ElevenLabs) give agents more choices for voice, image, and video generation. New image, video, and audio generation skills provide reusable workflows, and multi-turn image editing lets agents iteratively refine visuals across rounds. Plus: Nvidia NIM backend, quality rethinking subagent, and new CLI mode flags. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Spawn subagents for focused background work, including specialized roles. Personas, refinement, and dynamic evaluation criteria. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution — Structured plans with chunked execution, progress checkpoints, iterative review, and verification. Spec mode for formal requirements before execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Multiple providers (OpenAI, Google, ElevenLabs, Nano Banana 2, Grok Imagine) with multi-turn image editing and reusable media generation skills. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI & Web UI — Watch agents in real-time via Terminal UI with mode cycling, log analysis, tool batching, and keyboard shortcuts. Also available: browser-based Web UI. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.57, adding Delegated Subagent Protocol & Builder Subagent! 🚀 A new file-based delegation protocol enables container-to-host subagent spawning, and the builder subagent type handles large artifact generation with fresh context. Plus: updated Claude Code reasoning config, substantiveness tracking, and diagnostic report gating. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Agents can spawn subagents for focused work in the background, including specialized roles (builder, evaluator, explorer, critic, researcher, novelty). Both modes support personas, refinement, and dynamic evaluation criteria. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans with chunked execution, progress checkpoints, iterative review, and verification requirements. Spec mode for formal requirements before execution. Configurable plan depth and persistent storage for re-execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Native backend image routing and configurable video frame extraction. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents in real-time via Terminal UI with mode cycling (Normal/Planning/Execute/Analyzing), in-app log analysis, collapsible tool batching, and keyboard shortcuts. Also available: browser-based Web UI for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.56, adding Spec Plan Mode! 🚀 A new formal requirements specification workflow before execution with TUI spec mode support. Plus: critic subagent for quality assessment, targeted agent-to-agent messaging, media conversation continuity, and Codex OAuth login fix. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Both modes support personas, refinement, specialized subagent roles, and dynamic evaluation criteria. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans with chunked execution, progress checkpoints, iterative review, and verification requirements. Spec mode for formal requirements before execution. Configurable plan depth and persistent storage for re-execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Native backend image routing and configurable video frame extraction. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents in real-time via Terminal UI with mode cycling (Normal/Planning/Execute/Analyzing), in-app log analysis, collapsible tool batching, and keyboard shortcuts. Also available: browser-based Web UI for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

When someone you care about is stuck in a toxic environment, sometimes the kindest thing you can do is remind them - you see how hard this is for them, and they deserve peace. Sharing this for suffering friends. 🎵 youtu.be/-6UK3tmkCAA

We're excited to release MassGen v0.1.55, adding Specialized Subagent Types & Dynamic Evaluation Criteria! 🚀 Specialized subagent roles (evaluator, explorer, researcher, novelty) with a discovery-based type system via SUBAGENT.md frontmatter. Dynamic task-specific evaluation criteria with core/stretch gates replace static checklists. Plus: native backend routing for image understanding, configurable video frame extraction, and composition documentation. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel with shared workspaces, voting, and consensus — or decompose tasks into owned subtasks. Both modes support personas, refinement, specialized subagent roles, and dynamic evaluation criteria. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans with chunked execution, progress checkpoints, iterative review, and verification requirements. Configurable plan depth and persistent storage for re-execution. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — MCP tools and agent skills for file read/write (with worktree isolation), background tool execution, web browsing, computer use, and project editing — all with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. Native backend image routing and configurable video frame extraction. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents in real-time via Terminal UI with mode cycling (Normal/Planning/Execute/Analyzing), in-app log analysis, collapsible tool batching, and keyboard shortcuts. Also available: browser-based Web UI for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, execution traces, workspace history, cost tracking, and Logfire integration for full observability. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

I used the Orion AI agent (meetorion.app) to create a video demo of the A* search algorithm in the Russell & Norvig AI textbook - the 8 puzzle problem. Orion found the solution and created the video, also processes my new prompt without stopping the execution

One AI agent. One month. 110 tasks. 5 devices. 37 skills. → Attended meetings, speak & listen → Made autonomous phone calls → Handled emails, docs, calendar, msgs → Produced styled music videos & novels While I was working, flying, relaxing, dreaming... Dreamlife @MeetOrion

We're excited to release MassGen v0.1.54, adding Subagent Messaging & Copilot SDK Backend! 🚀 Send messages to running agents mid-execution to steer their work in real time — target specific agents or broadcast to all, with queued message management. Works for both main agents and subagents. Plus: new copilot backend powered by github-copilot-sdk, Gemini 3.1 Pro model support, and MCP hooks improvements. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access a wide range of open-source and proprietary models (local, Gemini 3.1 Pro, GPT-5.3-Codex, Claude 4.6 Opus/Sonnet, GLM-5, Grok 4.1, and other enterprise backends) and agents (Claude Code, Codex, GitHub Copilot). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel where they share workspaces, vote on the best solution, and reach consensus, or use decomposition mode to break complex tasks into subtasks owned by individual agents with a presenter synthesizing the final result. Both modes support unique personas, refinement, and subagent spawning. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans separating "what to build" from "how to build it", execute saved plans, and verify work at logical checkpoints with persistent storage and execution tracking. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — With MCP tools and agent skills, agents can read/write files, browse websites, connect to apps, use computers, and edit existing project files with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents collaborate in real-time through an interactive Textual Terminal UI (Normal/Planning/Execute via Shift+Tab), collapsible tool batching, timeline streaming, and keyboard shortcuts. Also available: browser-based Web UI with React frontend for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, workspace history, cost tracking, and optional Logfire integration make agent behavior easy to understand and debug. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

I asked 9 AI systems one factual question: how many of Pieter Abbeel's students are running startups? All 9 used live search. 7 got it wrong. Only 2 found the answer listed on Abbeel's own page. ChatGPT 5.2 Pro took 10m 42s. Orion took 50s. meetorion.app @MeetOrion

We're excited to release MassGen v0.1.53, focused on Background Tool Execution! 🚀 Background tool execution for non-blocking long-running work. Planning task verification requirements. TUI background job indicators and lifecycle controls. Subagent infrastructure groundwork with Evaluator and Explorer types. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access a wide range of open-source and proprietary models (local, Gemini 3, GPT-5.2, GPT-5.1-Codex, Claude, GLM, and other enterprise backends) and agents (AG2, AgentScope, Claude Code, LangGraph, and smolagents). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel where they share workspaces, vote on the best solution, and reach consensus, or use decomposition mode to break complex tasks into subtasks owned by individual agents with a presenter synthesizing the final result. Both modes support unique personas, refinement, and subagent spawning. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans separating "what to build" from "how to build it", execute saved plans, and verify work at logical checkpoints with persistent storage and execution tracking. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — With MCP tools and agent skills, agents can read/write files, browse websites, connect to apps, use computers, and edit existing project files with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents collaborate in real-time through an interactive Textual Terminal UI (Normal/Planning/Execute via Shift+Tab), collapsible tool batching, timeline streaming, and keyboard shortcuts. Also available: browser-based Web UI with React frontend for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, workspace history, cost tracking, and optional Logfire integration make agent behavior easy to understand and debug. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

We're excited to release MassGen v0.1.52, focused on Final Answer Modal & Coordination Quality Gates! 🚀 Dedicated final answer modal with tabbed answer and workspace/review interface. Substantive gate prevents low-value iteration rounds. Novelty injection combats premature convergence. Agent identity versioning for answer provenance tracking. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access a wide range of open-source and proprietary models (local, Gemini 3, GPT-5.2, GPT-5.1-Codex, Claude, GLM, and other enterprise backends) and agents (AG2, AgentScope, Claude Code, LangGraph, and smolagents). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel where they share workspaces, vote on the best solution, and reach consensus, or use decomposition mode to break complex tasks into subtasks owned by individual agents with a presenter synthesizing the final result. Both modes support unique personas, refinement, and subagent spawning. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans separating "what to build" from "how to build it", execute saved plans, and verify work at logical checkpoints with persistent storage and execution tracking. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — With MCP tools and agent skills, agents can read/write files, browse websites, connect to apps, use computers, and edit existing project files with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents collaborate in real-time through an interactive Textual Terminal UI (Normal/Planning/Execute via Shift+Tab), collapsible tool batching, timeline streaming, and keyboard shortcuts. Also available: browser-based Web UI with React frontend for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, workspace history, cost tracking, and optional Logfire integration make agent behavior easy to understand and debug. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…

"Write me a novel about a lighthouse keeper who discovers messages from the future." My AI built a 5-chapter outline, wrote Chapter 1, then I said "write the rest." 10 minutes later: 8,400 words, complete story. Bonus: Orion also did the recording below and deployed the novel online: orion-1771090137152-699.zinley.site meetorion.app @MeetOrion

We're excited to release MassGen v0.1.51, focused on Reviewing Coordination & Change Documents! 🚀 Review modal with multi-file diff visualization. Decision journal system for multi-agent coordination traceability. Changedoc-anchored evaluation checklists with gap reports. Drift conflict policy for safer change application. `--cwd-context` CLI flag. Release notes: github.com/massgen/MassGe… Feature highlights: 🧠Supports diverse backends — Access a wide range of open-source and proprietary models (local, Gemini 3, GPT-5.2, GPT-5.1-Codex, Claude, GLM, and other enterprise backends) and agents (AG2, AgentScope, Claude Code, LangGraph, and smolagents). 👉 More Info: docs.massgen.ai/en/latest/user… 🤖Build collaborative multi-agent teams — Run agents in parallel where they share workspaces, vote on the best solution, and reach consensus, or use decomposition mode to break complex tasks into subtasks owned by individual agents with a presenter synthesizing the final result. Both modes support unique personas, refinement, and subagent spawning. 👉 More Info: #multi-agent-coordination" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📋Task planning & execution— Agents create structured plans separating "what to build" from "how to build it", execute saved plans, and verify work at logical checkpoints with persistent storage and execution tracking. 👉 More Info: docs.massgen.ai/en/latest/user… 🛠️Work with files, apps, and projects — With MCP tools and agent skills, agents can read/write files, browse websites, connect to apps, use computers, and edit existing project files with a robust permission system. 👉 More Info: docs.massgen.ai/en/latest/user… 💾Session continuity — Multi-turn conversations with context compression, session resumption, and evolving skills that persist and improve across sessions. 👉 More Info: #evolving-skills-code-mode" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 🖼️Multi-modal support — Agents can understand and generate files, videos, images, and audio collaboratively. 👉 More Info: docs.massgen.ai/en/latest/user… 🐳Code execution — Each agent can run commands (with Docker support, if installed) and use code mode for efficient filesystem-discoverable tools with large token reduction. 👉 More Info: docs.massgen.ai/en/latest/user… 🖥️Interactive TUI (Default) & Web UI — Watch agents collaborate in real-time through an interactive Textual Terminal UI (Normal/Planning/Execute via Shift+Tab), collapsible tool batching, timeline streaming, and keyboard shortcuts. Also available: browser-based Web UI with React frontend for visual workflows. 👉 More Info: docs.massgen.ai/en/latest/user… 🪪Trace workflows — Orchestration logs, workspace history, cost tracking, and optional Logfire integration make agent behavior easy to understand and debug. 👉 More Info: #logging-directory-structure" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user… 📤Session sharing — Export multi-turn sessions to GitHub Gist and view them in the MassGen Viewer web app with artifact previews. 👉 More Info: #sharing-sessions" target="_blank" rel="nofollow noopener">docs.massgen.ai/en/latest/user…