dsh-baton-mcp 0.6.8 → 0.6.9

package/README.md

@@ -53,7 +53,7 @@ Switched on, the plugin registers dsh-baton for Claude Code and Codex, for every
 
 That is all. To hand a task over, end the request with "hand it to DeepSeek Harness". Updating, removing, Windows, and registering by hand when the plugin could not, are [further down](#install-details).
 
-**New in 0.6.4:** Codex waits in one `dsh_wait` call of up to 59 minutes, so a run that ends within the hour costs one model request to wait for, and the result arrives as an ordinary reply in the chat. Update the plugin in Harness and new Codex conversations have it, with no restart.
+**Installed on the day of a release?** For about a day after a release, Harness installs the previous version, and its install dialog says so. Open the plugin's page and press **Update to x.y.z** (更新到), or add the plugin with its version, such as `dsh-baton-host@0.6.8`.
 
 <p align="center"><picture><source media="(prefers-color-scheme: dark)" srcset="./assets/readme/en/divider-dark.svg"><img src="./assets/readme/en/divider.svg" width="100%" alt=""></picture></p>
 
@@ -284,7 +284,7 @@ When you update or remove the plugin, come from dsh-bridge, run from source, or
 
 **`npx -y dsh-baton-mcp setup`** does the same registration from a terminal: for a plugin that could not register, or when you come from an MCP directory. With the plugin already registered it changes nothing, then points you to the plugin if it is missing.
 
-**Updating.** Harness has an update button only for its official plugins, so: Plugins → dsh-baton-host → **Uninstall**, then **Add plugin** → `dsh-baton-host`. For a while after a release, Harness installs the previous version instead (pnpm's `minimumReleaseAge` holds back versions published very recently, and the install dialog says so); to get the new one at once, type the version: `dsh-baton-host@0.6.4`. A pre-release must always be added by its exact version, such as `dsh-baton-host@0.6.0-next.0`: Harness resolves `@next` to a broken placeholder. Plain `dsh-baton-host` gets the latest release. Switched on again, the plugin brings the MCP server to the same version; new conversations use it. If the two are ever out of step, the first tool call that notices it says so, naming the older side and the newer side's exact version to install.
+**Updating.** On Harness 0.2.2 and later: Plugins → dsh-baton-host → **Update to x.y.z** (更新到) under Version; if it still says up to date, press **Check again** (重新检查) first. Harness 0.2.1 has no update button for this plugin: **Uninstall**, then **Add plugin** → `dsh-baton-host`. For about a day after a release, Harness installs the previous version instead (pnpm's `minimumReleaseAge` holds back versions published very recently, and the install dialog says so); to get the new one at once, type the version: `dsh-baton-host@0.6.8`. A pre-release must always be added by its exact version, such as `dsh-baton-host@0.6.0-next.0`: Harness resolves `@next` to a broken placeholder. Plain `dsh-baton-host` gets the latest release. Switched on again, the plugin brings the MCP server to the same version; new conversations use it. If the two are ever out of step, the first tool call that notices it says so, naming the older side and the newer side's exact version to install.
 
 **Removing.** Harness → Plugins → dsh-baton-host → **Uninstall**, then `npx -y dsh-baton-mcp setup --remove`. The plugin cannot tell an uninstall from Harness quitting, so it leaves the MCP entries to that command, which also deletes `~/.dsh-baton/app`; run records stay in `~/.dsh-baton` until you delete it.
 

package/README.zh-CN.md

@@ -53,7 +53,7 @@ dsh-baton 是一个 MCP 服务器加一个 Harness 小插件。主 Agent 用一
 
 就这些。要转交任务，在需求最后加一句“交给 DeepSeek Harness 去做”。更新、卸载、Windows，以及插件没能注册时怎么手动注册，在[后面](#安装细节)。
 
-**0.6.4 新变化**：Codex 一次 `dsh_wait` 最多等 59 分钟，一小时以内跑完的任务等待只花一次模型请求，结果作为普通回复回到对话里。在 Harness 里更新插件，新开的 Codex 对话就用上了，不用重启。
+**刚发版就安装？** 新版本发布后大约一天内，Harness 会装上一个版本，安装弹窗里会说明。打开插件详情页点 **更新到 x.y.z**，或者添加插件时带上版本号，比如 `dsh-baton-host@0.6.8`。
 
 <p align="center"><picture><source media="(prefers-color-scheme: dark)" srcset="./assets/readme/zh/divider-dark.svg"><img src="./assets/readme/zh/divider.svg" width="100%" alt=""></picture></p>
 
@@ -284,7 +284,7 @@ Harness 插件连不上时：
 
 **`npx -y dsh-baton-mcp setup`** 在终端里做同样的注册：用于插件没能注册的情况，或者从 MCP 目录过来的用户。插件已经注册过时它什么都不改；插件没装的话，它会提示去添加。
 
-**更新**。Harness 只给官方插件提供"更新"按钮，所以要：插件 → dsh-baton-host → **卸载**，再**添加插件** → `dsh-baton-host`。新版本刚发布的一段时间里，Harness 会装上一个版本（pnpm 的 `minimumReleaseAge` 会暂缓刚发布的版本，安装弹窗里会说明）；想马上用新版，输入带版本号的名字：`dsh-baton-host@0.6.4`。预发布版必须写确切版本号添加，比如 `dsh-baton-host@0.6.0-next.0`：Harness 会把 `@next` 解析成一个坏掉的占位包。只输入 `dsh-baton-host` 装的是最新正式版。重新启用后，插件会把 MCP 服务器也更新到同一版本，新开的对话就会用上。万一两边版本对不上，第一次发现不一致的工具调用会提示，并写明哪一边旧、另一边的确切版本号。
+**更新**。Harness 0.2.2 及以上：插件 → dsh-baton-host → 版本一栏里的 **更新到 x.y.z**；如果还显示已是最新版本，先点 **重新检查**。Harness 0.2.1 没有这个插件的更新按钮：**卸载**，再**添加插件** → `dsh-baton-host`。新版本发布后大约一天内，Harness 会装上一个版本（pnpm 的 `minimumReleaseAge` 会暂缓刚发布的版本，安装弹窗里会说明）；想马上用新版，输入带版本号的名字：`dsh-baton-host@0.6.8`。预发布版必须写确切版本号添加，比如 `dsh-baton-host@0.6.0-next.0`：Harness 会把 `@next` 解析成一个坏掉的占位包。只输入 `dsh-baton-host` 装的是最新正式版。重新启用后，插件会把 MCP 服务器也更新到同一版本，新开的对话就会用上。万一两边版本对不上，第一次发现不一致的工具调用会提示，并写明哪一边旧、另一边的确切版本号。
 
 **卸载**。Harness → 插件 → dsh-baton-host → **卸载**，再运行 `npx -y dsh-baton-mcp setup --remove`。插件分不清"被卸载"和"Harness 退出"，所以 MCP 注册交给这条命令删，它也会删掉 `~/.dsh-baton/app`；运行记录留在 `~/.dsh-baton`，不需要的话自己删掉。
 

package/host/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-baton-host",
-  "version": "0.6.8",
+  "version": "0.6.9",
   "description": "DeepSeek Harness plugin for dsh-baton: hand coding tasks from Claude Code or Codex to DeepSeek Harness on your machine. Adding it is the whole install: it registers the dsh-baton MCP server for Claude Code and Codex, runs each task live in the Harness window, wakes the main agent when the run ends and hands back the changed files, the project's own test results and a requirement-by-requirement checklist.",
   "type": "module",
   "main": "lib/index.js",
@@ -20,7 +20,7 @@
     "LICENSE"
   ],
   "dependencies": {
-    "dsh-baton-mcp": "0.6.8"
+    "dsh-baton-mcp": "0.6.9"
   },
   "license": "MIT",
   "dsh": {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-baton-mcp",
-  "version": "0.6.8",
+  "version": "0.6.9",
   "mcpName": "io.github.JialaoLiu/dsh-baton",
   "description": "Pass the baton from Claude Code, Codex or any MCP client to a local DeepSeek Harness: an MCP server that hands off tasks, continues a specific session, shows the run live in Harness and hands the result back when it is done.",
   "type": "module",

package/server.json

@@ -7,12 +7,12 @@
     "url": "https://github.com/JialaoLiu/dsh-baton",
     "source": "github"
   },
-  "version": "0.6.8",
+  "version": "0.6.9",
   "packages": [
     {
       "registryType": "npm",
       "identifier": "dsh-baton-mcp",
-      "version": "0.6.8",
+      "version": "0.6.9",
       "transport": {
         "type": "stdio"
       }

package/src/cli.mjs

@@ -91,6 +91,7 @@ async function main() {
       const total_s = Number(flags.timeout) || 24 * 3600
       const waitStarted = Date.now()
       const st = await core.api.wait({ run_id: rest[0], timeout_s: total_s })
+      const stop = core.stopHint(st)
       // A Harness goal goes on after the run's own round: so does this wait, until the goal
       // stops (or the next report, with --until turn), with every round since the run began.
       if (core.goalGoesOn(st)) {
@@ -113,6 +114,10 @@ async function main() {
         // The hint points at evidence: it goes with it (files changed, checks run, other runs in the folder).
         ...(st.evidence ? { evidence: st.evidence } : {}),
         ...(st.state === 'completed' && !st.decided_for_user ? { hint: core.endedHint(st) } : {}),
+        // Stopped by the user, cancelled, or cut off by a Harness restart: say so, or the client
+        // takes it for a failure in the work and sends it again.
+        ...(st.harness_down ? { harness_down: true, hint: core.HARNESS_DOWN_HINT.replace('call dsh_wait again with the same run_id', 'run this same wait command again') } : {}),
+        ...(stop ? { ...(st.turn_end ? { turn_end: st.turn_end } : {}), ...stop } : {}),
         ...(st.needs_input ? { needs_input: st.needs_input, question: st.question, ...(st.question ? { relay: core.relayText(st.question) } : {}), hint: st.question ? 'The run asks the user something. If the user allowed you to decide, choose by your own judgment and answer with dsh_answer (or dsh-baton answer), passing reason, and tell the user what you chose and why. Otherwise show the user relay, written in the user\'s language, with its first line filled in with your own pick and a reason from your own judgment of the question and the task (harness_suggested is only the asker\'s pick, never a reason by itself), and ask the user first. Then wait again.' : 'The run waits for a person in the Harness UI. Tell the user, then wait again.' } : {}) })
       process.exitCode = st.state === 'completed' ? 0 : st.needs_input ? 3 : 1
       return

package/src/core.mjs

@@ -147,7 +147,8 @@ function hostCallAt(endpoint, method, params = {}, { timeoutMs = 15000 } = {}) {
       msg.error ? finish(new Error(String(msg.error))) : finish(null, msg.result)
     })
     conn.on('error', (e) => finish(Object.assign(e, { hostDown: true })))
-    conn.on('close', () => finish(new Error(`dsh-baton-host closed the connection before answering ${method}`)))
+    // Harness quitting closes the connection mid-call: marked so a wait can wait for it to come back.
+    conn.on('close', () => finish(Object.assign(new Error(`dsh-baton-host closed the connection before answering ${method}`), { hostClosed: true })))
   })
 }
 
@@ -623,12 +624,33 @@ const NEEDS_INPUT_HINT = (st) => st.question
   ? `The run stopped to ask the user ${st.question.count} question(s) (in question). Unless the user let you decide, show the user relay, written in the user's language, with its first line filled in with your own pick and reason (not the asker's pick and not its option description repeated), then ask which they want. ${ASKED} After answering, wait for the run again.`
   : 'The run waits for something only a person can do in the Harness UI (an approval, for example). Tell the user, then wait for the run again.'
 
+// A run the user stopped in the Harness UI: its turn ended aborted, by the user. With the usual
+// hint, Codex took that for a failure and sent the task again each time the user stopped it
+// (Windows, 2026-11-08). The UI stop leaves background jobs of the session running; dsh_cancel
+// stops them too.
+export const stoppedByUser = (st) => st?.state === 'failed' && st.turn_end?.kind === 'aborted' && st.turn_end?.reason?.kind === 'user'
+export const USER_STOPPED_HINT = 'The user stopped this run in Harness (stopped_by_user). It is not a failure to repair: do not send it again, continue it or send fixes. Tell the user it was stopped and what evidence shows it got done, then ask what to do next. Background jobs it started may still be running in that session; stop them with dsh_cancel and its session_id only if the user asks.'
+// Ended by something other than the work: Harness restarting under the run, or a cancel (dsh_cancel,
+// or in Harness: the user stopping the session drops its queued messages, or removing one). Sending
+// again on one's own would override the user here too.
+export const INTERRUPTED_HINT = 'Harness restarted or quit while this run was going, and its turn was stopped (interrupted). It is not a failure in the work: do not send it again or continue it on your own. Tell the user Harness was restarted and what evidence shows the run got done, then ask whether to run it again or continue in the same session.'
+export const CANCELLED_HINT = 'This run was cancelled before it finished: with dsh_cancel, or in Harness, where the user stopped the session or removed the queued message. If you did not cancel it yourself, the user did: do not send it again. Tell the user it was cancelled and ask what to do next.'
+// The hint for a run that did not end by its own work, or null for one that did.
+export function stopHint(st) {
+  if (stoppedByUser(st)) return { stopped_by_user: true, hint: USER_STOPPED_HINT }
+  if (st?.state === 'interrupted') return { hint: INTERRUPTED_HINT }
+  if (st?.state === 'cancelled') return { hint: CANCELLED_HINT }
+  return null
+}
+
 // The result of a run that ended, kept short: the client already knows what it sent, and a
 // completed run's event log is only the work behind final_text (dsh_status still has it).
 // A run that did not complete keeps its last events, to show where it stopped.
 function ended({ recent_events, message_preview, backend, ...st }) {
   const out = { ...st, ...(st.state !== 'completed' && recent_events?.length ? { recent_events: recent_events.slice(-5) } : {}) }
   if (outOfBalance(st)) return { ...out, hint: QUOTA_HINT }
+  const stop = stopHint(st)
+  if (stop) return { ...out, ...stop }
   // decided_for_user first, so a client that skims to final_text still meets it.
   return st.decided_for_user
     ? { hint: `${DECIDED_HINT} ${endedHint(st)}`, decided_for_user: st.decided_for_user, ...out }
@@ -757,6 +779,10 @@ export async function followThrough(params, { total_s, chunk_s = 300, onChunk, s
   }
 }
 
+// How often a wait looks for Harness again while it is not running.
+const HOST_RETRY_MS = Number(process.env.DSH_BATON_HOST_RETRY_MS) || 2000
+export const HARNESS_DOWN_HINT = 'DeepSeek Harness is not running (closed, or restarting), so the run cannot be followed right now; Harness ends a running turn when it quits. Do not send the task again. Tell the user Harness is closed and to open it; then call dsh_wait again with the same run_id: it reports how the run ended (interrupted, if Harness quit while it ran).'
+
 export const api = {
   async workspaces() {
     if (!(await useHost())) throw new Error('Listing workspaces needs the dsh-baton-host plugin running inside DeepSeek Harness.')
@@ -828,8 +854,20 @@ export const api = {
     return changed ? { ...st, changed_files: changed } : st
   },
   async wait({ run_id, timeout_s = 300 }) {
-    if (isHostRun(run_id)) return withEvidence(shaped(tag('host', await hostCall('wait', { run_id, timeout_s }, { timeoutMs: (timeout_s + 15) * 1000 }))))
-    return withEvidence(tag('headless', await waitRun(run_id, { timeout_s })))
+    if (!isHostRun(run_id)) return withEvidence(tag('headless', await waitRun(run_id, { timeout_s })))
+    // Harness quit or restarting: wait for it to come back within the same call (the plugin then
+    // reports the run as interrupted) instead of failing, which had Codex retry every few seconds.
+    const deadline = Date.now() + timeout_s * 1000
+    for (;;) {
+      const left = Math.max(1, Math.ceil((deadline - Date.now()) / 1000))
+      try {
+        return withEvidence(shaped(tag('host', await hostCall('wait', { run_id, timeout_s: left }, { timeoutMs: (left + 15) * 1000 }))))
+      } catch (error) {
+        if (!(error?.hostDown || error?.hostClosed)) throw error
+        if (Date.now() + HOST_RETRY_MS >= deadline) return { run_id, state: 'unknown', timed_out: true, harness_down: true, backend: 'host' }
+        await new Promise((r) => setTimeout(r, HOST_RETRY_MS))
+      }
+    }
   },
   async cancel({ run_id, session_id }) {
     if (run_id && !isHostRun(run_id)) return tag('headless', cancelRun(run_id))

package/src/mcp.mjs

@@ -122,6 +122,7 @@ async function waitInChunks({ run_id }, progress, signal, meta) {
     // Never spin: a chunk that came back early still takes a moment before the next one.
     if (Date.now() - t0 < 500) await new Promise((r) => setTimeout(r, 500))
     if (Date.now() >= deadline) {
+      if (st.harness_down) return { run_id, state: st.state, timed_out: true, harness_down: true, hint: core.HARNESS_DOWN_HINT }
       return {
         run_id, state: st.state, timed_out: true,
         ...(st.waiting_on ? { waiting_on: st.waiting_on } : {}),
@@ -182,6 +183,7 @@ async function waitAnyInChunks(run_ids, progress, signal, meta) {
     if (Date.now() - t0 < 500) await new Promise((r) => setTimeout(r, 500))
     if (Date.now() >= deadline) {
       const asking = running.filter((r) => r.state === 'waiting').map((r) => r.run_id)
+      if ([...answers.values()].some((st) => st.harness_down)) return { run_ids, timed_out: true, harness_down: true, still_running: running, hint: `${core.HARNESS_DOWN_HINT.replace('the same run_id', 'the same run_ids')}` }
       return {
         run_ids, timed_out: true, still_running: running,
         hint: `None of these runs has ended yet. Call dsh_wait again with the same run_ids, with no progress check or progress note in between (work the user asked you to do meanwhile is fine).${asking.length ? ` ${asking.join(', ')} still wait${asking.length === 1 ? 's' : ''} for a person (see dsh_status for the question): answer with dsh_answer if you may, or tell the user.` : ''}`,
@@ -200,12 +202,15 @@ function anyResult({ done, running }) {
       : res
   })
   const ids = running.map((r) => r.run_id)
+  // Stopped by the user, cancelled or cut off by a Harness restart: not work to send back.
+  const stopped = ended.filter((r) => r.stopped_by_user || r.state === 'interrupted' || r.state === 'cancelled').map((r) => r.run_id)
+  const leave = stopped.length ? ` ${stopped.join(', ')} did not end by ${stopped.length === 1 ? 'its' : 'their'} own work (stopped by the user, cancelled or cut off by a Harness restart, see ${stopped.length === 1 ? 'its' : 'each'} hint): send ${stopped.length === 1 ? 'it' : 'them'} nothing more on your own and leave ${stopped.length === 1 ? 'it' : 'them'} out of the next wait.` : ''
   const next = ids.length
     ? `Then call dsh_wait again with run_ids: the ones still running (${ids.join(', ')}) plus the run_id of every fix you sent, so whichever ends next comes back first. Do not wait for one run on its own while others are going.`
     : 'None of these runs is still going.'
   return {
     ended, still_running: running,
-    hint: `${ended.length} of the runs ${ended.length === 1 ? 'has' : 'have'} ended or stopped to ask (ended; each has its own hint). Deal with ${ended.length === 1 ? 'it' : 'them'} now: review against what you asked, and send any fixes with dsh_send to the same session_id right away, or answer a question. ${next}`,
+    hint: `${ended.length} of the runs ${ended.length === 1 ? 'has' : 'have'} ended or stopped to ask (ended; each has its own hint). Deal with ${ended.length === 1 ? 'it' : 'them'} now: review against what you asked, and send any fixes with dsh_send to the same session_id right away, or answer a question.${leave} ${next}`,
   }
 }