[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-92588":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":15,"stars30d":16,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":17,"rankGlobal":9,"rankLanguage":9,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":9,"pushedAt":9,"updatedAt":23,"readmeContent":24,"aiSummary":25,"trendingCount":15,"starSnapshotCount":15,"syncStatus":14,"lastSyncTime":26,"discoverSource":27},92588,"damon-ade","per-simmons\u002Fdamon-ade","per-simmons","ADE — an agentic development environment for macOS. A roster of persistent coding agents, each with its own repo, runtime CLI, and memory.",null,"TypeScript",74,32,52,2,0,22,46.76,"Other",false,"main",true,[],"2026-07-22 04:02:06","# ADE\n\nAn agentic development environment for macOS. ADE is a local-first, single-user desktop app where you build a roster of persistent coding agents and work alongside them in the terminal. Every agent is a durable identity — its own name, photo, git repository, runtime CLI, and long-lived memory — not a throwaway chat session. You come back to the same agent tomorrow and it remembers what it learned today.\n\nThe interface is a two-level left rail. **Teams** group your work (a name and a square photo); inside each team live **Agents** (a name and a circular photo). Selecting an agent opens its workspace: a strip of **session** tabs, each a real terminal running the agent's coding CLI inside that agent's own git worktree. A **model bar** under the tabs lets you spawn a session on a different model without leaving the agent. On the right, the **Agent Files** panel shows the agent's memory growing as it works.\n\nADE runs whatever CLI coding agents you already have installed. Claude Code, OpenAI's Codex, and OpenCode are first-class runtimes. The model bar can also launch sessions on Kimi K2.7, MiniMax M3, and GLM 5.2 through a single OpenRouter key you enter once, in-app. Nothing here is a hosted service — your code, your keys, and your agents' memory all stay on your machine.\n\n## Screenshots\n\n\u003C!-- TODO: rail with teams + agents -->\n\u003C!-- TODO: agent workspace with session tabs + model bar -->\n\u003C!-- TODO: Agent Files panel showing memory -->\n\n## Install\n\n### Download (recommended)\n\nDownload the signed DMG from the [latest release](https:\u002F\u002Fgithub.com\u002Fper-simmons\u002Fdamon-ade\u002Freleases\u002Flatest), open it, and drag ADE to your Applications folder. macOS only.\n\n\n### Build from source\n\nRequires [Bun](https:\u002F\u002Fbun.sh) 1.0+.\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fper-simmons\u002Fdamon-ade.git\ncd REPO\nbun install\ncd apps\u002Fdesktop\nbun run compile:app        # builds main + preload + renderer into dist\u002F\nbunx electron .            # launches the built app\n```\n\n`compile:app` runs the full production build; `bunx electron .` then launches it directly. (Avoid `electron-vite preview` for a full run — it can exhaust memory.)\n\n## Prerequisites\n\nADE orchestrates coding CLIs; it does not bundle them. You need:\n\n- **Git** — required. Each agent gets its own repository or worktree. Install Apple's command line tools with `xcode-select --install`.\n- **At least one agent CLI.** Claude Code is recommended, because it also powers the Kimi, MiniMax, and GLM sessions from the model bar (they run Claude Code pointed at OpenRouter):\n\n  ```bash\n  npm i -g @anthropic-ai\u002Fclaude-code\n  ```\n\n  Optionally add the other runtimes:\n\n  ```bash\n  npm i -g @openai\u002Fcodex        # OpenAI GPT-5.5 sessions\n  npm i -g opencode-ai          # OpenCode runtime\n  ```\n\n- **Node.js** — only as the vehicle for installing the CLIs above via `npm`. ADE itself does not need a separate Node runtime.\n- **An OpenRouter API key** — only if you want the open-model sessions (Kimi K2.7, MiniMax M3, GLM 5.2). You enter it in-app the first time you launch one of those models; see the walkthrough. The Claude and OpenAI runtimes authenticate through their own CLIs (your Anthropic and ChatGPT\u002FOpenAI logins) and need no key here.\n\n## Walkthrough\n\n**1. First launch.** ADE opens on a start screen with a single action: **Create a team**. There are no agents until a team exists, so start here.\n\n**2. Create a team.** Give it a name (for example, `Newsletter`). Optionally click the square photo thumbnail to pick an image — teams are the top level of the rail, so a photo makes them easy to find at a glance. The team appears in the left rail.\n\n**3. Create an agent.** Hover the team's header in the rail and click the **+** button (\"New agent\"). In the New Agent dialog:\n   - **Name** — required (for example, `Scout`).\n   - **Role** — optional. A sentence describing what this agent is for. Leave it blank if you'd rather shape the agent by talking to it — ADE seeds the agent's identity file either way, and it refines itself over time.\n   - **Runtime** — the coding CLI this agent runs: **Claude**, **Codex**, or **OpenCode**. Claude is the default.\n   - **Repository** — start a new empty repo, clone from a URL, or point at an existing local path.\n\n   ADE creates the agent, gives it its own git worktree, and scaffolds its memory in the background.\n\n**4. Add profile photos.** Right-click any agent in the rail and choose **Change Photo** (or **Remove Photo**) to give it a circular avatar. Team photos are set the same way from the team's header menu. Photos are optional but make a busy rail readable.\n\n**5. Sessions start automatically.** Opening an agent that has no sessions yet automatically spawns one — a terminal tab running the agent's runtime CLI in its worktree. That's the agent, live. Open more session tabs whenever you want parallel threads of work.\n\n**6. Switch models from the model bar.** Below the session tabs is a quiet row of model logos: **Claude** (the default), **OpenAI** (Codex on GPT-5.5), **Kimi K2.7**, **MiniMax M3**, and **GLM 5.2**. Click any logo to open a new session in the current agent's worktree running that model — the same code, a different model, no context switch.\n\n**7. Connect OpenRouter (first open model only).** The first time you click Kimi, MiniMax, or GLM, ADE asks for your OpenRouter API key (get one at [openrouter.ai\u002Fkeys](https:\u002F\u002Fopenrouter.ai\u002Fkeys)). Paste it and choose **Save & Launch**. The key is encrypted with the macOS keychain, stored locally, and injected only into the agent's terminal — it never leaves your machine and is never shown back to the app's UI. You enter it once; later open-model sessions launch straight away.\n\n**8. Watch the memory grow.** The **Agent Files** panel on the right lists the agent's memory surface, grouped into **Memory**, **Skills**, and **Worktree**. It starts nearly empty and fills in as the agent learns — its identity, your profile, its notes, and any skills it writes for itself. Click a file to open it in a viewer tab.\n\n## How memory works\n\nEvery ADE agent keeps a persistent, self-curated memory, adapted from the [Hermes agent](https:\u002F\u002Fgithub.com\u002FNousResearch\u002Fhermes-agent). The design is deliberately simple: plain markdown files the agent reads at the start of every session and writes back to as it learns. The files live outside the git worktree, so they survive branch and worktree churn and are never committed to your code.\n\nEach agent's memory is a small set of files:\n\n- **AGENT.md** — a short identity and operating brief (who the agent is, its role, its standing preferences).\n- **USER.md** — a profile of you: name, preferences, communication style, hard rules.\n- **MEMORY.md** — the agent's own notes: project conventions, tool quirks, lessons learned, plus an index into any longer topic files.\n- **Skills** — reusable, multi-step procedures the agent writes for itself, each a `SKILL.md` whose body loads only when relevant.\n\nA write-back protocol travels with the memory, telling the agent when to save (a stated preference, a correction, a durable fact), when to skip (trivia, one-off state, anything easily re-discovered), and to consolidate rather than endlessly append. A session-end reflection loop prompts the agent to review the conversation and update its memory and skills before it finishes, so the next session starts smarter. On Claude Code this reflection is enforced by a native stop hook; on the other runtimes it runs by convention at session boundaries.\n\nThe same canonical files feed every runtime through thin, auto-generated bridge files — a `CLAUDE.md`, an OpenCode config, or a regenerated Codex `AGENTS.md` — so you can switch an agent's runtime without losing its memory. See [docs\u002Fmemory.md](docs\u002Fmemory.md) for the full design.\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) and our [Code of Conduct](CODE_OF_CONDUCT.md).\n\n## License\n\nADE is a modified derivative of [Superset](https:\u002F\u002Fgithub.com\u002Fsuperset-sh\u002Fsuperset) (Copyright Superset, Inc.), distributed under the **Elastic License 2.0** — see [LICENSE.md](LICENSE.md). Third-party dependency notices are in [THIRD-PARTY-NOTICES.md](THIRD-PARTY-NOTICES.md). The agent memory architecture is adapted from [NousResearch\u002Fhermes-agent](https:\u002F\u002Fgithub.com\u002FNousResearch\u002Fhermes-agent) (MIT).\n","ADE 是一个面向 macOS 的本地化智能编程代理开发环境，支持用户创建并管理多个持久化编码代理。每个代理拥有独立的 Git 仓库、专属终端 CLI、长期记忆及可切换的 AI 模型（如 Claude Code、Codex、OpenCode、Kimi、GLM 等），所有数据与密钥均保留在本地。其双栏界面支持按团队组织代理，每个代理提供多会话终端标签页和可视化记忆文件面板，强调身份持久性与隐私可控性。适用于个人开发者在本地构建、调试和协作式 AI 编程工作流，尤其适合需长期上下文记忆、模型灵活切换及代码资产隔离的场景。","2026-07-09 02:30:25","CREATED_QUERY"]