- name
SKILL.md
技能文件保留在发布者的仓库中。
在终端中粘贴复制的命令。它会把这个 SKILL.md 安装到对应的智能体中。
npx skills add amplifthq/openmeld --skill verify-service-connection -a cursor -y在终端中粘贴复制的命令。它会把这个 SKILL.md 安装到对应的智能体中。
npx skills add amplifthq/openmeld --skill verify-service-connection -a claude-code -y在终端中粘贴复制的命令。它会把这个 SKILL.md 安装到对应的智能体中。
npx skills add amplifthq/openmeld --skill verify-service-connection -a codex -y在终端中粘贴复制的命令。它会把这个 SKILL.md 安装到对应的智能体中。
npx skills add amplifthq/openmeld --skill verify-service-connection -a grok -y在 Grok Bot 聊天中发送复制的消息。Bot 会从 GitHub 获取这个 SKILL.md,并添加到你所有 Bot 共享的技能中。
Install this skill: https://github.com/amplifthq/openmeld/tree/HEAD/.agents/skills/verify-service-connection来自发布者
Only the local Service's online state reports healthy. Connecting, starting, and authentication problems report degraded, and stopping reports idle. The loop model also checks that quitting prevents reconnection. Core owns the separate Wakeability projection.
This is an ordering question about one connection loop, so the spec is TLA+ and the checker is TLC 1.7.4 (jar sha256 936a262061c914694dfd669a543be24573c45d5aa0ff20a8b96b23d01e050e88). The code check calls toHealthProjection, classifyConnectionFailureReason, and createServiceConnectionStateTracker. Java 11 or newer must be on PATH. The check downloads the jar into ~/.cache/openmeld-tlc/ on first run.
Run it when a change touches any of these:
clients/openmeld-cli/src/local-service/state/service-connection-state.ts
createServiceConnectionStateTrackertoHealthProjectionclassifyConnectionFailureReasonclients/openmeld-cli/src/local-service/carrier/service-runner.ts
connecting, online, and the failure statesnoteAuthenticationRequiredStatenoteReconnectFailureStatemarkStoppingConfirmed against commit 965d9eb3219fa2c35789f9ab9a2206ab8699de9b. Move that commit forward only after you have confirmed that ServiceConnection.tla and check.mjs still describe those functions.
From the repository root:
node .agents/skills/verify-service-connection/check.mjs
Exit 0 means the health projection, the upgrade-rejection classification, and the loop model all hold. pnpm install is not required for this check.
Read every TLC result with /statechart. The user-facing labels are in view.mjs next to the spec.
toHealthProjection for all six service connection states. In particular, connecting must report degraded/suspect until the connection is online.a non-401 stream upgrade rejection is internal: classifyConnectionFailureReason treated a 500, or a message that merely contains "401" or "socket", as sign-in or network. That sends the user to the wrong recovery.NoReconnectAfterQuit failed: the model now has a path from stopping back to connecting. Check markStopping and the while (!stopRequested) loop before you weaken the rule.connectionState.set in the run loop, or the health read that forces stopping while draining.One service process. Backoff durations are not in the model. sign_in_unavailable is degraded reason auth_unavailable, not authentication_required. The chart hides "you quit" from connecting, sign-in, and the reconnecting states. You can still quit from those; the chart shows quit from Starting and Connected, and every other state can still try again or fail.
Use /maintain-formal-verification-skill to bring this check back in line.
Check the local OpenMeld Service connection loop. The computer is shown as connected only while the session is online, stopping is idle, and quitting does not reconnect. Use when changing the service
---name: verify-service-connectiondescription: Check the local OpenMeld Service connection loop. The computer is shown as connected only while the session is…---# .agents/skills/verify-service-connection/SKILL.md概览
Check the local OpenMeld Service connection loop. The computer is shown as connected only while the session is online, stopping is idle, and quitting does not reconnect. Use when changing the service run loop, toHealthProjection, or classifyConnectionFailureReason.
常见问题
从发布者处下载 .agents/skills/verify-service-connection/SKILL.md,或复制路径后打开仓库。PlugMCP 只收录该技能,不托管文件。
发布者将它存放在 https://github.com/amplifthq/openmeld。
Claude、Cursor、Codex 和 Grok Build 可以读取你从发布者仓库添加的 SKILL.md。在聊天中发送技能的 GitHub 文件夹链接,Grok Bot 就会安装它。