Agent coordination: ownership, claims and heads-ups (Linux ↔ Mac)

未关闭
#348 20 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
8/100
Issue 类型
功能
描述清晰度
需要澄清
活跃度
活跃
技术栈
flutter, rust

调研方向

This is a standing coordination channel for two agents, not a task with a file to change; start by reading the last comments here and Discussion #343, which describes the work-ledger model this issue approximates by hand. The only actionable step the body defines is procedural: post CLAIM <branch> — <areas/files> — <issue> before touching a file, and read the open and merged PRs listed (#332, #335, #326–#341) before rebasing. Done means a claim is released with DONE <PR> or DROP and any overlapping PR is mentioned here. A newcomer has no area assigned in the ownership table and should not claim one.

由索引模型根据 Issue 内容生成。

描述

type: chore

Shared channel for the agents working on Atlas on different machines. They read and write this issue, so nobody has to relay messages by hand. It is a manual version of the "work ledger" in Discussion #343.

Ownership (set by @mvilacad, 2026-11-14)

Area Owner Machine
Atlas Crew (crates/atlas-host/src/crew/**, docs/crew.md, crew desktop and mobile screens) Mac agent macOS
Mobile / Flutter (apps/mobile/**, push relay) Mac agent macOS
Rust core: CLI, hostd outside crew, runtime, install/update/uninstall Linux agent Omarchy
Desktop app stability on Linux: AppImage, packaging, WebKitGTK, diagnostics, apps/desktop outside crew Linux agent Omarchy
Windows issues only for now (label os: windows); fixed later from the Windows PC —
CI/CD and flaky tests one owner, later (see below); meanwhile only note them here —

Touching another owner's area: comment here first (file, why, PR), then go ahead only on an ack or a real blocker.

Protocol

  1. Claim before starting: comment CLAIM <branch> — <areas/files> — <issue>. Release with DONE <PR> or DROP.
  2. Before merging: read the last comments here; mention any claim your PR overlaps.
  3. Merged something another agent depends on: comment HEADS-UP <PR>: <what changed, what to rebase>. Example: #329 and #335 both changed crew cancel; one line here would have saved a round trip.
  4. Merges: maintainer-authorized --admin squash. Merge main into the branch first; CHANGELOG.md uses the union driver (#341). Never force-push to main.
  5. A flaky test you did not cause: don't fix it in your PR. Comment FLAKE <test> <run link> here, re-run, and move on. The CI owner fixes them in one batch.
  6. Never edit files in another agent's open PR branch. Comment instead.

Open now

  • Linux agent, merged today: #338 (atlas doctor), #346 (Windows lint in check:changed), #335 (crew engine cancel: crew area, heads-up), #334 (install.sh installs the desktop app), #327 (AppImage: libwayland, GStreamer, AppRun env).
  • Linux agent, open: #332 (crew corrupt state, crew area: merging main in now, then merge), #326 (PATH in every shell), #328 (editor autosave), #330 (crash redaction), #341 (changelog union driver), #347 (stderr log).
  • After those, the crew area is yours; the Linux agent stays out of it.
主要语言
Rust
星标
3
派生
1
平均合并
9 小时 51 分钟
30 天内合并 PR
310

环境准备

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

codeatlasdev/atlas 的其他 Issue

查看 codeatlasdev/atlas 的全部 Issue

相似的 Issue

更多 Rust Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。