geo-ai-search-optimization 1.2.9 → 1.2.11

package/README.md

@@ -103,6 +103,48 @@ geo-ai-search-optimization agent-runbook ./reports/apply-plan.json --json
 - 回报清单
 - 可直接复制给 agent 的 runbook prompt
 
+## Agent Executor 命令
+
+如果你希望 agent 不只是拿到 runbook，而是直接得到“这一轮先执行哪 1 个任务”的入口，可以直接用 `agent-executor`：
+
+```bash
+geo-ai-search-optimization agent-executor ./your-site
+geo-ai-search-optimization agent-executor ./reports/apply-plan.json
+geo-ai-search-optimization agent-executor ./reports/apply-plan.json --task fix-02 --json
+```
+
+它会输出：
+
+- 当前先做哪一包
+- 为什么先做这一包
+- do-now checklist
+- stop checklist
+- success checklist
+- 验证命令
+- 给用户的回报模板
+- 可直接复制给 agent 的 executor prompt
+
+## Agent Batch Executor 命令
+
+如果你希望 agent 不只做 1 包，而是把前 2 到 3 包排成一个连续执行序列，同时仍然保持“一次只推进一包”，可以直接用 `agent-batch-executor`：
+
+```bash
+geo-ai-search-optimization agent-batch-executor ./your-site
+geo-ai-search-optimization agent-batch-executor ./reports/apply-plan.json
+geo-ai-search-optimization agent-batch-executor ./reports/apply-plan.json --task fix-02 --count 3 --format json --out ./reports/agent-batch-executor.json
+```
+
+`agent-batch-executor` 会输出：
+
+- 当前批次先推进哪几包
+- 为什么按这个顺序推进
+- 每一包的 do-now checklist
+- 每一包的 stop checklist
+- 每一包的 success checklist
+- 批次总验证命令
+- 批次收尾命令
+- 可直接复制给 agent 的 batch prompt
+
 ## Quick Start
 
 如果你要从 0 到 1 启动一个 GEO 项目，建议照这个顺序做。
@@ -499,6 +541,8 @@ geo-ai-search-optimization install --target ./tmp/custom-skills --json
 geo-ai-search-optimization auto-flow "audit this site and tell me the next skill"
 geo-ai-search-optimization agent-session ./your-site
 geo-ai-search-optimization agent-runbook ./your-site
+geo-ai-search-optimization agent-executor ./your-site
+geo-ai-search-optimization agent-batch-executor ./your-site
 geo-ai-search-optimization skills
 geo-ai-search-optimization where
 geo-ai-search-optimization doctor
@@ -562,6 +606,20 @@ geo-ai-search-optimization help
 - 输出开始前检查、工作规则、停止条件、验证清单、回报清单
 - 新增 `geo-ai-search-optimization-agent-runbook` skill
 
+## New in 1.2.10
+
+- 新增 `agent-executor` 命令
+- 把 runbook 再收敛成“这一轮先做哪 1 包”的单任务入口
+- 输出 do-now checklist、stop checklist、success checklist、验证命令和回报模板
+- 新增 `geo-ai-search-optimization-agent-executor` skill
+
+## New in 1.2.11
+
+- 新增 `agent-batch-executor` 命令
+- 让 agent 可以连续推进前几包任务，但仍然保持一次只执行一包
+- 输出批次顺序、每包检查清单、批次总验证命令和收尾命令
+- 新增 `geo-ai-search-optimization-agent-batch-executor` skill
+
 ## New in 1.2.5
 
 - 新增 `publish-pack`
@@ -751,6 +809,8 @@ The installed package now includes a bundled GEO skill pack, including:
 - `geo-ai-search-optimization-auto-flow`
 - `geo-ai-search-optimization-agent-session`
 - `geo-ai-search-optimization-agent-runbook`
+- `geo-ai-search-optimization-agent-executor`
+- `geo-ai-search-optimization-agent-batch-executor`
 - `geo-ai-search-optimization-usage`
 - `geo-ai-search-optimization-agent-handoff`
 - `geo-ai-search-optimization-repair-loop`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "geo-ai-search-optimization",
-  "version": "1.2.9",
+  "version": "1.2.11",
   "description": "Install and run a Generative Engine Optimization (GEO)-first, SEO-supported Codex skill for website optimization.",
   "type": "module",
   "bin": {

package/resources/geo-ai-search-optimization/references/skill-bundle-map.md

@@ -52,6 +52,26 @@ Best for:
 - telling the next agent what to check before acting
 - defining stop conditions, validation checklist, and reporting checklist
 
+### `geo-ai-search-optimization-agent-executor`
+
+Use this when the next agent should stop planning and start one concrete execution packet.
+
+Best for:
+
+- selecting the first packet to execute now
+- reducing a runbook into one actionable task
+- giving the next agent a single-task prompt, validation commands, and reply template
+
+### `geo-ai-search-optimization-agent-batch-executor`
+
+Use this when the next agent should continuously advance the first few execution packets, but still only work on one packet at a time.
+
+Best for:
+
+- turning an `apply-plan` into a short execution queue
+- making the next agent finish packet 1 before packet 2 starts
+- giving the next agent a batch-level prompt, per-packet checklists, and final closeout commands
+
 ## Usage guide
 
 ### `geo-ai-search-optimization-usage`

package/resources/geo-ai-search-optimization-agent-batch-executor/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: geo-ai-search-optimization-agent-batch-executor
+description: Turn a GEO input into a short multi-packet execution queue for the next agent. Use when an agent should continuously advance the first 2 to 3 GEO packets in order, while still executing only one packet at a time, with per-packet checklists, validation commands, and closeout steps.
+---
+
+# GEO Agent Batch Executor
+
+Use this skill when the next agent should not stop at one task, but also should not expand into an uncontrolled multi-task run.
+
+`GEO = Generative Engine Optimization`
+
+## What it does
+
+- select the first few execution packets from an apply-plan or equivalent artifact
+- keep the execution order explicit
+- enforce one-packet-at-a-time progress
+- provide per-packet do-now, stop, and success checklists
+- give the next agent one batch prompt plus final closeout commands
+
+## Best use
+
+- when `agent-executor` is too narrow because the next agent should keep going after packet 1
+- when a PM wants “finish the first 2 to 3 packets in order”
+- when the next agent needs a short queue, not a full open-ended plan

package/resources/geo-ai-search-optimization-agent-batch-executor/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "GEO Agent Batch Executor"
+  short_description: "Queue the first GEO packets for one-by-one execution"
+  default_prompt: "Use $geo-ai-search-optimization-agent-batch-executor to choose the first 2 to 3 GEO packets to advance in order, while still executing only one packet at a time."

package/resources/geo-ai-search-optimization-agent-executor/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: geo-ai-search-optimization-agent-executor
+description: Turn a GEO input into a single-task execution entrypoint for the next agent. Use when an agent should stop planning and start one concrete GEO packet now, with a selected task, do-now checklist, validation commands, success criteria, and a reply template.
+---
+
+# GEO Agent Executor
+
+Use this skill when the next agent should execute one packet now, not keep expanding the plan.
+
+`GEO = Generative Engine Optimization`
+
+## What it does
+
+- select the first or requested execution packet
+- compress the runbook into one task entrypoint
+- define do-now, stop, and success checklists
+- provide validation commands and a reply template
+- make the next agent focus on one packet at a time
+
+## Best use
+
+- when `agent-runbook` is still too broad
+- when a PM or another agent asks “this round, what should we do first?”
+- when execution should start from one concrete packet

package/resources/geo-ai-search-optimization-agent-executor/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "GEO Agent Executor"
+  short_description: "Start one GEO execution packet right now"
+  default_prompt: "Use $geo-ai-search-optimization-agent-executor to choose the first GEO packet to execute now, with a do-now checklist, validation commands, and a reply template."

package/resources/geo-ai-search-optimization-usage/SKILL.md

@@ -13,24 +13,26 @@ Treat this tool as a PM-friendly GEO workflow for websites.
 
 `GEO = Generative Engine Optimization`
 
-The package is best explained as sixteen layers:
+The package is best explained as eighteen layers:
 
 1. `auto-flow`: auto-select the next skill and command chain
 2. `agent-session`: build a runnable session for the next agent
 3. `agent-runbook`: execution manual and checklist for the next agent
-4. `skills`: inspect the bundled skill package
-5. `onboard-url` / `onboard`: first look
-6. `scan`: raw signal check
-7. `audit` / `report`: diagnosis
-8. `fix-plan` / `owner-board`: execution planning
-9. `agent-handoff`: agent takeover package
-10. `apply-plan`: execution loop
-11. `completion-report`: closeout
-12. `handoff-bundle`: all-in-one package
-13. `share-pack`: audience-ready delivery
-14. `export-pack`: folder export
-15. `html-pack` / `publish-pack`: browsable and final delivery output
-16. `pm-brief` / `roadmap`: stakeholder alignment
+4. `agent-executor`: choose one packet to execute right now
+5. `agent-batch-executor`: queue the first few packets, but still advance one packet at a time
+6. `skills`: inspect the bundled skill package
+7. `onboard-url` / `onboard`: first look
+8. `scan`: raw signal check
+9. `audit` / `report`: diagnosis
+10. `fix-plan` / `owner-board`: execution planning
+11. `agent-handoff`: agent takeover package
+12. `apply-plan`: execution loop
+13. `completion-report`: closeout
+14. `handoff-bundle`: all-in-one package
+15. `share-pack`: audience-ready delivery
+16. `export-pack`: folder export
+17. `html-pack` / `publish-pack`: browsable and final delivery output
+18. `pm-brief` / `roadmap`: stakeholder alignment
 
 ## Recommended command order
 
@@ -40,6 +42,8 @@ If the user only has a website URL:
 npx geo-ai-search-optimization auto-flow https://example.com
 npx geo-ai-search-optimization agent-session https://example.com
 npx geo-ai-search-optimization agent-runbook https://example.com
+npx geo-ai-search-optimization agent-executor https://example.com
+npx geo-ai-search-optimization agent-batch-executor https://example.com
 npx geo-ai-search-optimization onboard-url https://example.com
 npx geo-ai-search-optimization pm-brief https://example.com
 npx geo-ai-search-optimization roadmap https://example.com
@@ -51,6 +55,8 @@ If the user has the website codebase:
 npx geo-ai-search-optimization auto-flow ./your-site
 npx geo-ai-search-optimization agent-session ./your-site
 npx geo-ai-search-optimization agent-runbook ./your-site
+npx geo-ai-search-optimization agent-executor ./your-site
+npx geo-ai-search-optimization agent-batch-executor ./your-site
 npx geo-ai-search-optimization scan ./your-site
 npx geo-ai-search-optimization audit ./your-site
 npx geo-ai-search-optimization fix-plan ./your-site
@@ -71,6 +77,8 @@ npx geo-ai-search-optimization roadmap ./your-site
 - `auto-flow`: auto-select the next skill and command order from a task brief, URL, project path, or GEO artifact
 - `agent-session`: build a step-by-step session packet for the next agent from the same kinds of inputs
 - `agent-runbook`: build a checklist-driven runbook with preflight, validation, and reporting rules
+- `agent-executor`: select one packet to execute now and package it into a single-task entrypoint
+- `agent-batch-executor`: line up the first few packets in execution order while preserving one-packet-at-a-time discipline
 - `onboard-url`: first-time website check from a live URL
 - `onboard`: interactive first-time onboarding
 - `skills`: list the bundled skills and decide which skill or command chain fits the task
@@ -98,6 +106,8 @@ When explaining the tool to a user:
 - if the user or the next agent is unsure where to start, move them to `auto-flow` first
 - if the user wants something the next agent can follow step by step, move them to `agent-session`
 - if the user wants the next agent to follow a checklist and execution manual, move them to `agent-runbook`
+- if the user wants the next agent to start one concrete task now, move them to `agent-executor`
+- if the user wants the next agent to continuously advance the first 2 to 3 packets in order, move them to `agent-batch-executor`
 - explain the result in PM language, not implementation jargon
 - if the user sounds new, start with `onboard-url` or `quick-start`
 - if the user wants another agent to take over, move them to `agent-handoff`

package/src/agent-batch-executor.js

@@ -0,0 +1,327 @@
+import { readFile } from "node:fs/promises";
+import { createApplyPlan } from "./apply-plan.js";
+import { createAgentExecutor } from "./agent-executor.js";
+import { writeScanOutput } from "./scan.js";
+
+const VALID_FORMATS = new Set(["markdown", "json"]);
+
+function normalizeFormat(format) {
+  const resolved = (format || "markdown").toLowerCase();
+  if (!VALID_FORMATS.has(resolved)) {
+    throw new Error(`不支持的 agent-batch-executor 格式：${format}。可选值：${Array.from(VALID_FORMATS).join(", ")}`);
+  }
+  return resolved;
+}
+
+function normalizeCount(count) {
+  if (count == null) {
+    return 3;
+  }
+
+  const parsed = Number.parseInt(count, 10);
+  if (!Number.isInteger(parsed) || parsed <= 0) {
+    throw new Error("--count 必须是正整数");
+  }
+
+  return Math.min(parsed, 5);
+}
+
+async function loadApplyPlanArtifact(input) {
+  const raw = await readFile(input, "utf8");
+  const parsed = JSON.parse(raw);
+
+  if (parsed?.kind === "geo-apply-plan") {
+    return parsed;
+  }
+  if (parsed?.kind === "geo-agent-runbook" && parsed.applyPlan?.kind === "geo-apply-plan") {
+    return parsed.applyPlan;
+  }
+  if (parsed?.kind === "geo-agent-executor" && parsed.applyPlan?.kind === "geo-apply-plan") {
+    return parsed.applyPlan;
+  }
+
+  return null;
+}
+
+async function resolveApplyPlan(input) {
+  const artifact = await loadApplyPlanArtifact(input).catch(() => null);
+  if (artifact) {
+    return artifact;
+  }
+
+  return createApplyPlan(input, {
+    format: "json"
+  });
+}
+
+function selectPackets(applyPlan, count, startTaskId) {
+  const packets = applyPlan?.packets || [];
+  if (packets.length === 0) {
+    return [];
+  }
+
+  let startIndex = 0;
+  if (startTaskId) {
+    startIndex = packets.findIndex((packet) => packet.id === startTaskId);
+    if (startIndex === -1) {
+      throw new Error(`找不到执行包：${startTaskId}`);
+    }
+  }
+
+  return packets.slice(startIndex, startIndex + count);
+}
+
+async function buildPacketExecutors(input, packets, options) {
+  const results = [];
+
+  for (const packet of packets) {
+    const executor = await createAgentExecutor(input, {
+      format: "json",
+      intent: options.intent,
+      taskId: packet.id
+    });
+
+    results.push({
+      position: results.length + 1,
+      id: packet.id,
+      title: packet.title,
+      priority: packet.priority,
+      owner: packet.owner,
+      executorMode: executor.executorMode,
+      doNowChecklist: executor.doNowChecklist,
+      stopChecklist: executor.stopChecklist,
+      successChecklist: executor.successChecklist,
+      validationCommands: executor.validationCommands,
+      userReplyTemplate: executor.userReplyTemplate,
+      executorPrompt: executor.executorPrompt,
+      startCommand: `geo-ai-search-optimization agent-executor ${executor.source} --task ${packet.id}`,
+      nextCommands: executor.nextCommands
+    });
+  }
+
+  return results;
+}
+
+function inferBatchMode(packetExecutors) {
+  if (packetExecutors.length === 0) {
+    return "needs-context";
+  }
+
+  const modes = new Set(packetExecutors.map((packet) => packet.executorMode));
+  if (modes.size === 1) {
+    return packetExecutors[0].executorMode;
+  }
+  return "mixed";
+}
+
+function buildSelectionReason(packets, applyPlan, count, startTaskId) {
+  if (packets.length === 0) {
+    return "当前还没有可批量执行的任务包，说明需要先补上下文或先生成 apply-plan。";
+  }
+
+  if (startTaskId) {
+    return `已从指定任务 ${startTaskId} 开始，连续选出后续 ${packets.length} 包，按顺序推进。`;
+  }
+
+  if (packets.length === 1) {
+    return "当前只拿到 1 个任务包，因此直接把它作为本轮批处理入口。";
+  }
+
+  const available = applyPlan?.packets?.length || packets.length;
+  if (available < count) {
+    return `计划最多取前 ${count} 包，但当前只有 ${available} 包可用，因此按现有顺序推进这 ${packets.length} 包。`;
+  }
+
+  return `已按优先级选出前 ${packets.length} 包，按顺序推进，但一次只执行一包。`;
+}
+
+function buildBatchRules(batchMode) {
+  const items = [
+    "始终一次只推进一包，不要同时展开多个修复包。",
+    "每做完一包，都先检查完成标准和验证命令，再决定是否进入下一包。",
+    "如果上一包没有通过验收，不要直接跳到下一包。"
+  ];
+
+  if (batchMode === "direct-fix") {
+    items.push("每包都优先改复用层，而不是只 patch 单页。");
+  }
+  if (batchMode === "recommendation-only") {
+    items.push("当前更适合给建议，不要伪造已经完成的代码修改。");
+  }
+
+  return items;
+}
+
+function buildQueueSummary(packetExecutors) {
+  return packetExecutors.map((packet, index) => ({
+    position: index + 1,
+    id: packet.id,
+    title: packet.title,
+    priority: packet.priority,
+    owner: packet.owner,
+    startCommand: packet.startCommand,
+    stopIf: packet.stopChecklist[packet.stopChecklist.length - 1] || "当前包未达成完成标准时不要继续。",
+    afterSuccess:
+      index < packetExecutors.length - 1
+        ? `进入下一包：${packetExecutors[index + 1].id}`
+        : "进入 completion-report 与 handoff-bundle 收尾。"
+  }));
+}
+
+function buildValidationRollup(packetExecutors) {
+  const commands = new Set();
+  for (const packet of packetExecutors) {
+    for (const command of packet.validationCommands) {
+      commands.add(command);
+    }
+  }
+  return Array.from(commands);
+}
+
+function buildFinalCommands(source) {
+  return [
+    `geo-ai-search-optimization completion-report ${source}`,
+    `geo-ai-search-optimization handoff-bundle ${source}`,
+    `geo-ai-search-optimization publish-pack ${source}`
+  ];
+}
+
+function buildBatchPrompt(batch) {
+  const lines = [
+    "你现在进入 GEO 多任务批次执行模式。",
+    `当前输入：${batch.source}`,
+    `批次模式：${batch.batchMode}`,
+    `批次目标：${batch.goal}`,
+    `批次选择原因：${batch.selectionReason}`,
+    "请遵守：一次只推进一包，上一包通过验证后再进入下一包。"
+  ];
+
+  if (batch.packetExecutors.length > 0) {
+    lines.push("当前批次顺序：");
+    for (const packet of batch.packetExecutors) {
+      lines.push(`${packet.position}. ${packet.id}｜${packet.title}`);
+    }
+  } else {
+    lines.push("当前还没有可执行包。请先告诉用户缺什么上下文。");
+  }
+
+  lines.push("每一包都要输出：先检查什么、怎么做、如何验证、完成后如何回报。");
+  return lines.join("\n");
+}
+
+export async function createAgentBatchExecutor(input, options = {}) {
+  const format = normalizeFormat(options.format);
+  const count = normalizeCount(options.count);
+  const applyPlan = await resolveApplyPlan(input);
+  const selectedPackets = selectPackets(applyPlan, count, options.taskId);
+  const packetExecutors = await buildPacketExecutors(input, selectedPackets, options);
+  const batchMode = inferBatchMode(packetExecutors);
+  const source = packetExecutors[0]?.startCommand ? applyPlan.source || input : applyPlan?.source || input;
+
+  return {
+    kind: "geo-agent-batch-executor",
+    input,
+    source,
+    sourceType: applyPlan?.sourceType || "unknown",
+    artifactKind: applyPlan?.kind || "unknown",
+    format,
+    batchMode,
+    packetCount: packetExecutors.length,
+    goal:
+      packetExecutors.length > 0
+        ? `按顺序推进这 ${packetExecutors.length} 包，但每次只执行一包。`
+        : "当前先补上下文，再进入多任务批次执行。",
+    selectionReason: buildSelectionReason(selectedPackets, applyPlan, count, options.taskId),
+    batchRules: buildBatchRules(batchMode),
+    queueSummary: buildQueueSummary(packetExecutors),
+    packetExecutors,
+    validationRollup: buildValidationRollup(packetExecutors),
+    finalCommands: buildFinalCommands(source),
+    batchPrompt: buildBatchPrompt({
+      source,
+      batchMode,
+      goal:
+        packetExecutors.length > 0
+          ? `按顺序推进这 ${packetExecutors.length} 包，但每次只执行一包。`
+          : "当前先补上下文，再进入多任务批次执行。",
+      selectionReason: buildSelectionReason(selectedPackets, applyPlan, count, options.taskId),
+      packetExecutors
+    }),
+    applyPlan
+  };
+}
+
+export function renderAgentBatchExecutorMarkdown(batch) {
+  const lines = [
+    "# GEO Agent Batch Executor",
+    "",
+    `- 输入：\`${batch.source}\``,
+    `- 来源类型：\`${batch.sourceType}\``,
+    `- 工件类型：\`${batch.artifactKind}\``,
+    `- 批次模式：\`${batch.batchMode}\``,
+    `- 批次数量：\`${batch.packetCount}\``,
+    `- 目标：${batch.goal}`,
+    `- 选择原因：${batch.selectionReason}`,
+    "",
+    "## 批次规则",
+    ""
+  ];
+
+  for (const item of batch.batchRules) {
+    lines.push(`- [ ] ${item}`);
+  }
+
+  lines.push("", "## 执行顺序", "");
+  for (const item of batch.queueSummary) {
+    lines.push(`### ${item.position}. ${item.id}｜${item.title}`, "");
+    lines.push(`- 优先级：${item.priority}`);
+    lines.push(`- Owner：${item.owner}`);
+    lines.push(`- 开始命令：\`${item.startCommand}\``);
+    lines.push(`- 继续条件：${item.afterSuccess}`);
+    lines.push(`- 停下条件：${item.stopIf}`);
+    lines.push("");
+  }
+
+  for (const packet of batch.packetExecutors) {
+    lines.push(`## ${packet.position}. ${packet.id}｜${packet.title}`, "");
+    lines.push("- Do now：");
+    for (const item of packet.doNowChecklist) {
+      lines.push(`  - [ ] ${item}`);
+    }
+    lines.push("- 停下条件：");
+    for (const item of packet.stopChecklist) {
+      lines.push(`  - [ ] ${item}`);
+    }
+    lines.push("- 完成标准：");
+    for (const item of packet.successChecklist) {
+      lines.push(`  - [ ] ${item}`);
+    }
+    lines.push("- 验证命令：");
+    for (const command of packet.validationCommands) {
+      lines.push(`  - \`${command}\``);
+    }
+    lines.push("- 回报模板：");
+    lines.push("```text");
+    lines.push(packet.userReplyTemplate);
+    lines.push("```");
+    lines.push("");
+  }
+
+  lines.push("## 批次总验证命令", "");
+  for (const command of batch.validationRollup) {
+    lines.push(`- \`${command}\``);
+  }
+
+  lines.push("", "## 最终收尾命令", "");
+  for (const command of batch.finalCommands) {
+    lines.push(`- \`${command}\``);
+  }
+
+  lines.push("", "## 可直接复制给 Agent 的 Batch Prompt", "", "```text", batch.batchPrompt, "```");
+
+  return `${lines.join("\n")}\n`;
+}
+
+export async function writeAgentBatchExecutorOutput(outputPath, content) {
+  return writeScanOutput(outputPath, content);
+}

package/src/agent-executor.js

@@ -0,0 +1,319 @@
+import { readFile } from "node:fs/promises";
+import { createApplyPlan } from "./apply-plan.js";
+import { createAgentRunbook } from "./agent-runbook.js";
+import { writeScanOutput } from "./scan.js";
+
+const VALID_FORMATS = new Set(["markdown", "json"]);
+
+function normalizeFormat(format) {
+  const resolved = (format || "markdown").toLowerCase();
+  if (!VALID_FORMATS.has(resolved)) {
+    throw new Error(`不支持的 agent-executor 格式：${format}。可选值：${Array.from(VALID_FORMATS).join(", ")}`);
+  }
+  return resolved;
+}
+
+async function loadJsonArtifact(input) {
+  const raw = await readFile(input, "utf8");
+  return JSON.parse(raw);
+}
+
+async function resolveRunbookAndPlan(input, options = {}) {
+  const runbook = await createAgentRunbook(input, {
+    format: "json",
+    intent: options.intent,
+    taskId: options.taskId
+  });
+
+  if (runbook.applyPlan?.kind === "geo-apply-plan") {
+    return {
+      runbook,
+      applyPlan: runbook.applyPlan
+    };
+  }
+
+  if (runbook.sourceType === "json") {
+    const parsed = await loadJsonArtifact(input).catch(() => null);
+    if (parsed?.kind === "geo-apply-plan") {
+      return {
+        runbook,
+        applyPlan: parsed
+      };
+    }
+    if (parsed?.kind === "geo-agent-runbook" && parsed.applyPlan?.kind === "geo-apply-plan") {
+      return {
+        runbook,
+        applyPlan: parsed.applyPlan
+      };
+    }
+  }
+
+  try {
+    const applyPlan = await createApplyPlan(input, {
+      format: "json",
+      taskId: options.taskId
+    });
+    return {
+      runbook,
+      applyPlan
+    };
+  } catch {
+    return {
+      runbook,
+      applyPlan: null
+    };
+  }
+}
+
+function inferExecutorMode(runbook, applyPlan) {
+  if (!applyPlan?.packets?.length) {
+    if (runbook.contextNeeded?.length) {
+      return "needs-context";
+    }
+    return "planning-only";
+  }
+  return applyPlan.executionType || "guided-planning";
+}
+
+function selectPacket(applyPlan, taskId) {
+  if (!applyPlan?.packets?.length) {
+    return null;
+  }
+
+  if (taskId) {
+    const selected = applyPlan.packets.find((packet) => packet.id === taskId);
+    if (!selected) {
+      throw new Error(`找不到执行包：${taskId}`);
+    }
+    return selected;
+  }
+
+  return applyPlan.packets[0];
+}
+
+function buildSelectionReason(packet, applyPlan, taskId) {
+  if (!packet) {
+    return "当前还没有可执行任务包，说明需要先补上下文或先生成 apply-plan。";
+  }
+  if (taskId) {
+    return `已按指定任务 ID 选中：${taskId}。`;
+  }
+  if (applyPlan?.selectedTaskCount > 1) {
+    return "默认选择第一包，代表当前优先级最高、最适合立刻开始的一项任务。";
+  }
+  return "当前只有一个任务包，因此直接将它作为本轮执行入口。";
+}
+
+function buildDoNowChecklist(packet, executorMode) {
+  if (!packet) {
+    return [];
+  }
+
+  const items = [
+    `先锁定任务范围：${packet.title}`,
+    ...packet.inspectTargets.slice(0, 3).map((item) => `先检查：${item}`),
+    ...packet.editPlan.slice(0, 3).map((item) => `执行动作：${item}`)
+  ];
+
+  if (executorMode === "direct-fix") {
+    items.push("开始修改前先确认会影响整类页面的复用层，而不是只 patch 单页。");
+  }
+
+  return items;
+}
+
+function buildStopChecklist(runbook, packet, executorMode) {
+  const items = [...(runbook.stopConditions || [])];
+
+  if (!packet) {
+    items.push("还没有拿到执行包时，不要假装已经进入真正修复阶段。");
+    return items;
+  }
+
+  if (executorMode === "recommendation-only") {
+    items.push("如果当前只有网址或建议上下文，不要输出伪造的代码修改结果。");
+  }
+
+  items.push("如果执行中发现这不是最高优先级任务，应先说明原因，再决定是否切换。");
+  return items;
+}
+
+function buildSuccessChecklist(packet) {
+  if (!packet) {
+    return [];
+  }
+
+  return [
+    ...packet.doneWhen,
+    "你已经能用 PM 可读的方式说明：改了什么、为什么改、验证结果是什么。"
+  ];
+}
+
+function buildNextCommands(runbook, packet, input) {
+  if (!packet) {
+    return runbook.nextCommands || [`geo-ai-search-optimization agent-runbook ${input}`];
+  }
+
+  const source = runbook.source;
+  return [
+    `geo-ai-search-optimization completion-report ${source}`,
+    `geo-ai-search-optimization handoff-bundle ${source}`
+  ];
+}
+
+function buildUserReplyTemplate(packet, executorMode) {
+  if (!packet) {
+    return [
+      "结论：我还不能直接执行，因为当前缺少足够上下文。",
+      "我目前确认到的情况：",
+      "- ",
+      "我还需要你补充：",
+      "- ",
+      "拿到这些之后，我会先执行：",
+      "- "
+    ].join("\n");
+  }
+
+  const closingLine =
+    executorMode === "direct-fix"
+      ? "我已经完成这一包可直接落地的部分，并附上复测结果。"
+      : executorMode === "recommendation-only"
+        ? "我已经完成这一包的建议版执行，但还需要更多上下文才能继续实际修改。"
+        : "我已经把这一包的执行顺序和阻塞项整理好，下一步可以继续推进。";
+
+  return [
+    `任务：${packet.title}`,
+    `结论：${closingLine}`,
+    "这次我先做了什么：",
+    "- ",
+    "我检查了哪些文件 / 页面 / 模板：",
+    "- ",
+    "我实际修改或建议了什么：",
+    "- ",
+    "验证结果：",
+    "- ",
+    "剩余风险或下一步：",
+    "- "
+  ].join("\n");
+}
+
+function buildExecutorPrompt(runbook, packet, executorMode, selectionReason) {
+  const lines = [
+    "你现在进入 GEO 单任务执行模式。",
+    `当前输入：${runbook.source}`,
+    `当前阶段：${runbook.stage}`,
+    `执行模式：${executorMode}`,
+    `任务选择原因：${selectionReason}`
+  ];
+
+  if (packet) {
+    lines.push(`本轮只做这一包：${packet.title}`);
+    lines.push("请你按这个顺序输出：");
+    lines.push("1. 先检查什么");
+    lines.push("2. 你准备如何执行这一包");
+    lines.push("3. 你会如何验证");
+    lines.push("4. 完成后如何向用户回报");
+  } else {
+    lines.push("当前还不能执行具体任务。请先明确缺什么上下文，再告诉用户下一步要补什么。");
+  }
+
+  lines.push("一次只推进一包，不要在同一轮同时展开多个任务。");
+  return lines.join("\n");
+}
+
+export async function createAgentExecutor(input, options = {}) {
+  const format = normalizeFormat(options.format);
+  const { runbook, applyPlan } = await resolveRunbookAndPlan(input, options);
+  const executorMode = inferExecutorMode(runbook, applyPlan);
+  const packet = selectPacket(applyPlan, options.taskId);
+  const selectionReason = buildSelectionReason(packet, applyPlan, options.taskId);
+
+  return {
+    kind: "geo-agent-executor",
+    input,
+    source: runbook.source,
+    sourceType: runbook.sourceType,
+    artifactKind: runbook.artifactKind,
+    format,
+    intent: runbook.intent,
+    stage: runbook.stage,
+    executorMode,
+    goal: packet ? `先完成这一包：${packet.title}` : runbook.goal,
+    selectionReason,
+    selectedSkill: {
+      name: "geo-ai-search-optimization-agent-executor",
+      displayName: "GEO Agent Executor",
+      shortDescription: "Start one GEO execution packet right now"
+    },
+    selectedPacket: packet,
+    doNowChecklist: buildDoNowChecklist(packet, executorMode),
+    stopChecklist: buildStopChecklist(runbook, packet, executorMode),
+    successChecklist: buildSuccessChecklist(packet),
+    validationCommands: packet?.validationCommands || runbook.validationChecklist || [],
+    userReplyTemplate: buildUserReplyTemplate(packet, executorMode),
+    nextCommands: buildNextCommands(runbook, packet, input),
+    executorPrompt: buildExecutorPrompt(runbook, packet, executorMode, selectionReason),
+    agentRunbook: runbook,
+    applyPlan
+  };
+}
+
+export function renderAgentExecutorMarkdown(executor) {
+  const lines = [
+    "# GEO Agent Executor",
+    "",
+    `- 输入：\`${executor.source}\``,
+    `- 输入类型：\`${executor.sourceType}\``,
+    `- 工件类型：\`${executor.artifactKind}\``,
+    `- 当前阶段：${executor.stage}`,
+    `- 执行模式：\`${executor.executorMode}\``,
+    `- 本轮目标：${executor.goal}`,
+    `- 任务选择原因：${executor.selectionReason}`,
+    "",
+    "## 现在先做什么",
+    ""
+  ];
+
+  for (const item of executor.doNowChecklist) {
+    lines.push(`- [ ] ${item}`);
+  }
+
+  if (executor.selectedPacket) {
+    lines.push("", "## 当前任务包", "");
+    lines.push(`- ID：\`${executor.selectedPacket.id}\``);
+    lines.push(`- 标题：${executor.selectedPacket.title}`);
+    lines.push(`- 优先级：${executor.selectedPacket.priority}`);
+    lines.push(`- Owner：${executor.selectedPacket.owner}`);
+    lines.push(`- 执行类型：\`${executor.selectedPacket.executionType}\``);
+    lines.push("");
+    lines.push("### 验证命令", "");
+    for (const command of executor.validationCommands) {
+      lines.push(`- \`${command}\``);
+    }
+  }
+
+  lines.push("", "## 需要停下来的情况", "");
+  for (const item of executor.stopChecklist) {
+    lines.push(`- [ ] ${item}`);
+  }
+
+  lines.push("", "## 完成标准", "");
+  for (const item of executor.successChecklist) {
+    lines.push(`- [ ] ${item}`);
+  }
+
+  lines.push("", "## 完成后怎么回报用户", "", "```text", executor.userReplyTemplate, "```");
+
+  lines.push("", "## 下一步推荐命令", "");
+  for (const command of executor.nextCommands) {
+    lines.push(`- \`${command}\``);
+  }
+
+  lines.push("", "## 可直接复制给 Agent 的 Executor Prompt", "", "```text", executor.executorPrompt, "```");
+
+  return `${lines.join("\n")}\n`;
+}
+
+export async function writeAgentExecutorOutput(outputPath, content) {
+  return writeScanOutput(outputPath, content);
+}

package/src/agent-session.js

@@ -49,6 +49,12 @@ function inferSkillForCommand(commandName, flow) {
   if (commandName === "agent-runbook") {
     return "geo-ai-search-optimization-agent-runbook";
   }
+  if (commandName === "agent-executor") {
+    return "geo-ai-search-optimization-agent-executor";
+  }
+  if (commandName === "agent-batch-executor") {
+    return "geo-ai-search-optimization-agent-batch-executor";
+  }
   if (commandName === "skills" || commandName === "quick-start") {
     return "geo-ai-search-optimization-usage";
   }
@@ -111,6 +117,10 @@ function inferStepPurpose(commandName, flow) {
       return "把当前结果交接成 agent 可继续执行的工件。";
     case "agent-runbook":
       return "把当前链路整理成 agent 可照着执行的手册和检查表。";
+    case "agent-executor":
+      return "把这一轮先做哪一个任务包压成单任务执行入口。";
+    case "agent-batch-executor":
+      return "把前几包任务排成连续执行顺序，但保持一次只推进一包。";
     case "apply-plan":
       return "把交接结果推进到具体执行包。";
     case "completion-report":
@@ -153,6 +163,10 @@ function inferExpectedArtifact(commandName) {
       return "agent 交接工件";
     case "agent-runbook":
       return "agent 执行手册";
+    case "agent-executor":
+      return "agent 单任务执行包";
+    case "agent-batch-executor":
+      return "agent 多任务批次执行包";
     case "apply-plan":
       return "执行包";
     case "completion-report":
@@ -188,6 +202,12 @@ function buildStepInstructions(parsedCommand, flow) {
   if (parsedCommand.commandName === "agent-runbook") {
     lines.push("先用 runbook 稳定顺序、检查项和回报方式，再开始真正执行。");
   }
+  if (parsedCommand.commandName === "agent-executor") {
+    lines.push("这一轮只推进一包任务，不要把多个修复包混在同一次执行里。");
+  }
+  if (parsedCommand.commandName === "agent-batch-executor") {
+    lines.push("批次模式下也要一次只做一包，上一包通过验证后再进入下一包。");
+  }
   if (parsedCommand.commandName === "agent-handoff" && flow.intent === "execute") {
     lines.push("如果还是 advice-only，说明还缺仓库或本地项目上下文。");
   }

package/src/auto-flow.js

@@ -44,6 +44,12 @@ function inferTaskTextMode(text) {
   if (/(share-pack|export-pack|html-pack|publish-pack|分享|导出|交付|外发|报告包)/i.test(normalized)) {
     return "share";
   }
+  if (/(batch-executor|batch executor|多包|前3包|批次执行|连续推进前几包)/i.test(normalized)) {
+    return "execute";
+  }
+  if (/(executor|先做哪一个|先做哪一包|single task|执行第一包|先执行一个任务)/i.test(normalized)) {
+    return "execute";
+  }
   if (/(runbook|checklist|playbook|执行手册|检查清单|操作手册)/i.test(normalized)) {
     return "execute";
   }
@@ -59,6 +65,10 @@ function inferTaskTextMode(text) {
   return "diagnose";
 }
 
+function taskTextWantsBatch(text) {
+  return /(batch-executor|batch executor|多包|前3包|前 3 包|批次执行|连续推进前几包|前几包)/i.test(String(text));
+}
+
 async function detectInput(input) {
   if (!input) {
     throw new Error("auto-flow 需要一个输入值，可以是任务描述、项目路径、网站网址或已导出的工件。");
@@ -150,9 +160,15 @@ function resolveEffectiveIntent(intent, detected) {
     return "share";
   }
   if (
-    ["geo-agent-handoff", "geo-agent-runbook", "geo-apply-plan", "geo-handoff-bundle", "geo-fix-plan"].includes(
-      detected.artifactKind
-    )
+    [
+      "geo-agent-handoff",
+      "geo-agent-runbook",
+      "geo-agent-executor",
+      "geo-agent-batch-executor",
+      "geo-apply-plan",
+      "geo-handoff-bundle",
+      "geo-fix-plan"
+    ].includes(detected.artifactKind)
   ) {
     return "execute";
   }
@@ -203,7 +219,8 @@ function buildCommandChain(detected, intent) {
         `geo-ai-search-optimization onboard-url ${source}`,
         `geo-ai-search-optimization fix-plan ${source}`,
         `geo-ai-search-optimization agent-handoff ${source}`,
-        `geo-ai-search-optimization agent-runbook ${source}`
+        `geo-ai-search-optimization agent-runbook ${source}`,
+        `geo-ai-search-optimization agent-executor ${source}`
       ];
     }
     return [
@@ -220,6 +237,7 @@ function buildCommandChain(detected, intent) {
         `geo-ai-search-optimization fix-plan ${source}`,
         `geo-ai-search-optimization agent-handoff ${source}`,
         `geo-ai-search-optimization agent-runbook ${source}`,
+        `geo-ai-search-optimization agent-executor ${source}`,
         `geo-ai-search-optimization apply-plan ${source}`
       ];
     }
@@ -238,6 +256,7 @@ function buildCommandChain(detected, intent) {
             `geo-ai-search-optimization fix-plan ${source}`,
             `geo-ai-search-optimization agent-handoff ${source}`,
             `geo-ai-search-optimization agent-runbook ${source}`,
+            `geo-ai-search-optimization agent-executor ${source}`,
             `geo-ai-search-optimization apply-plan ${source}`
           ]
         : [`geo-ai-search-optimization fix-plan ${source}`, `geo-ai-search-optimization owner-board ${source}`];
@@ -250,26 +269,41 @@ function buildCommandChain(detected, intent) {
       ];
     case "geo-fix-plan":
       return [
-        `geo-ai-search-optimization agent-handoff ${source}`,
         `geo-ai-search-optimization agent-runbook ${source}`,
-        `geo-ai-search-optimization apply-plan ${source}`
+        `geo-ai-search-optimization agent-executor ${source}`,
+        `geo-ai-search-optimization agent-batch-executor ${source}`,
+        `geo-ai-search-optimization completion-report ${source}`
       ];
     case "geo-agent-handoff":
       return [
         `geo-ai-search-optimization agent-runbook ${source}`,
-        `geo-ai-search-optimization apply-plan ${source}`,
+        `geo-ai-search-optimization agent-executor ${source}`,
+        `geo-ai-search-optimization agent-batch-executor ${source}`,
         `geo-ai-search-optimization completion-report ${source}`
       ];
     case "geo-agent-runbook":
       return [
-        `geo-ai-search-optimization apply-plan ${source}`,
+        `geo-ai-search-optimization agent-executor ${source}`,
+        `geo-ai-search-optimization agent-batch-executor ${source}`,
         `geo-ai-search-optimization completion-report ${source}`,
         `geo-ai-search-optimization handoff-bundle ${source}`
       ];
+    case "geo-agent-executor":
+      return [
+        `geo-ai-search-optimization completion-report ${source}`,
+        `geo-ai-search-optimization handoff-bundle ${source}`,
+        `geo-ai-search-optimization publish-pack ${source}`
+      ];
+    case "geo-agent-batch-executor":
+      return [
+        `geo-ai-search-optimization completion-report ${source}`,
+        `geo-ai-search-optimization handoff-bundle ${source}`,
+        `geo-ai-search-optimization publish-pack ${source}`
+      ];
     case "geo-apply-plan":
       return [
-        `geo-ai-search-optimization agent-runbook ${source}`,
-        `geo-ai-search-optimization apply-plan ${source}`,
+        `geo-ai-search-optimization agent-executor ${source}`,
+        `geo-ai-search-optimization agent-batch-executor ${source}`,
         `geo-ai-search-optimization completion-report ${source}`,
         `geo-ai-search-optimization handoff-bundle ${source}`
       ];
@@ -317,6 +351,9 @@ function pickSkillName(detected, intent) {
 
   switch (detected.artifactKind) {
     case "task-brief":
+      if (detected.inferredTextIntent === "execute" && taskTextWantsBatch(detected.source)) {
+        return "geo-ai-search-optimization-agent-batch-executor";
+      }
       return detected.inferredTextIntent === "share"
         ? "geo-ai-search-optimization-publish-pack"
         : detected.inferredTextIntent === "execute"
@@ -330,11 +367,13 @@ function pickSkillName(detected, intent) {
     case "geo-report:onboarding":
       return intent === "execute" ? "geo-ai-search-optimization-agent-handoff" : "geo-ai-search-optimization";
     case "geo-fix-plan":
-      return "geo-ai-search-optimization-agent-handoff";
     case "geo-agent-handoff":
     case "geo-agent-runbook":
+    case "geo-agent-executor":
     case "geo-apply-plan":
-      return "geo-ai-search-optimization-agent-runbook";
+      return "geo-ai-search-optimization-agent-executor";
+    case "geo-agent-batch-executor":
+      return "geo-ai-search-optimization-agent-batch-executor";
     case "geo-completion-report":
       return "geo-ai-search-optimization-completion-report";
     case "geo-handoff-bundle":
@@ -366,8 +405,19 @@ function buildSecondarySkillNames(primarySkill, intent, detected) {
   }
   if (
     intent === "execute" ||
-    ["geo-fix-plan", "geo-agent-handoff", "geo-agent-runbook", "geo-apply-plan"].includes(detected.artifactKind)
+    [
+      "geo-fix-plan",
+      "geo-agent-handoff",
+      "geo-agent-runbook",
+      "geo-agent-executor",
+      "geo-agent-batch-executor",
+      "geo-apply-plan"
+    ].includes(
+      detected.artifactKind
+    )
   ) {
+    names.add("geo-ai-search-optimization-agent-batch-executor");
+    names.add("geo-ai-search-optimization-agent-executor");
     names.add("geo-ai-search-optimization-agent-runbook");
     names.add("geo-ai-search-optimization-agent-handoff");
     names.add("geo-ai-search-optimization-repair-loop");
@@ -391,16 +441,28 @@ function buildStage(intent, detected) {
     return "执行复盘";
   }
   if (intent === "execute") {
-    return ["geo-fix-plan", "geo-agent-handoff", "geo-agent-runbook", "geo-apply-plan", "geo-handoff-bundle"].includes(
-      detected.artifactKind
-    )
+    return [
+      "geo-fix-plan",
+      "geo-agent-handoff",
+      "geo-agent-runbook",
+      "geo-agent-executor",
+      "geo-agent-batch-executor",
+      "geo-apply-plan",
+      "geo-handoff-bundle"
+    ].includes(detected.artifactKind)
       ? "Agent 执行"
       : "执行准备";
   }
   if (
-    ["geo-fix-plan", "geo-agent-handoff", "geo-agent-runbook", "geo-apply-plan", "geo-handoff-bundle"].includes(
-      detected.artifactKind
-    )
+    [
+      "geo-fix-plan",
+      "geo-agent-handoff",
+      "geo-agent-runbook",
+      "geo-agent-executor",
+      "geo-agent-batch-executor",
+      "geo-apply-plan",
+      "geo-handoff-bundle"
+    ].includes(detected.artifactKind)
   ) {
     return "Agent 执行";
   }

package/src/cli.js

@@ -2,6 +2,12 @@ import { fileURLToPath } from "node:url";
 import { readFile } from "node:fs/promises";
 import path from "node:path";
 import { createApplyPlan, renderApplyPlanMarkdown, writeApplyPlanOutput } from "./apply-plan.js";
+import {
+  createAgentBatchExecutor,
+  renderAgentBatchExecutorMarkdown,
+  writeAgentBatchExecutorOutput
+} from "./agent-batch-executor.js";
+import { createAgentExecutor, renderAgentExecutorMarkdown, writeAgentExecutorOutput } from "./agent-executor.js";
 import { createAgentHandoff, renderAgentHandoffMarkdown, writeAgentHandoffOutput } from "./agent-handoff.js";
 import { createAgentRunbook, renderAgentRunbookMarkdown, writeAgentRunbookOutput } from "./agent-runbook.js";
 import { createAgentSession, renderAgentSessionMarkdown, writeAgentSessionOutput } from "./agent-session.js";
@@ -63,6 +69,8 @@ function printHelp() {
       "  geo-ai-search-optimization auto-flow <input> [--intent <auto|diagnose|guide|execute|share|closeout>] [--json] [--out <file>]",
       "  geo-ai-search-optimization agent-session <input> [--intent <auto|diagnose|guide|execute|share|closeout>] [--json] [--out <file>]",
       "  geo-ai-search-optimization agent-runbook <input> [--intent <auto|diagnose|guide|execute|share|closeout>] [--task <id>] [--format <markdown|json>] [--out <file>]",
+      "  geo-ai-search-optimization agent-executor <input> [--intent <auto|diagnose|guide|execute|share|closeout>] [--task <id>] [--format <markdown|json>] [--out <file>]",
+      "  geo-ai-search-optimization agent-batch-executor <input> [--intent <auto|diagnose|guide|execute|share|closeout>] [--task <id>] [--count <count>] [--format <markdown|json>] [--out <file>]",
       "  geo-ai-search-optimization skills [--json]",
       "  geo-ai-search-optimization where",
       "  geo-ai-search-optimization doctor [--json]",
@@ -206,6 +214,61 @@ async function handleAgentRunbook(args) {
   process.stdout.write(renderedOutput);
 }
 
+async function handleAgentExecutor(args) {
+  const input = args.find((value) => !value.startsWith("-"));
+  if (!input) {
+    throw new Error("agent-executor 需要一个输入值，可以是任务描述、项目路径、网站网址或已导出的工件");
+  }
+
+  const format = getFlagValue(args, "--format");
+  const executor = await createAgentExecutor(input, {
+    intent: getFlagValue(args, "--intent"),
+    format,
+    taskId: getFlagValue(args, "--task")
+  });
+  const outputJson = format === "json";
+  const renderedOutput = outputJson
+    ? `${JSON.stringify(executor, null, 2)}\n`
+    : renderAgentExecutorMarkdown(executor);
+
+  const outputPath = getFlagValue(args, "--out");
+  if (outputPath) {
+    const resolvedOutputPath = await writeAgentExecutorOutput(outputPath, renderedOutput);
+    process.stdout.write(`已保存 agent executor：${resolvedOutputPath}\n`);
+    return;
+  }
+
+  process.stdout.write(renderedOutput);
+}
+
+async function handleAgentBatchExecutor(args) {
+  const input = args.find((value) => !value.startsWith("-"));
+  if (!input) {
+    throw new Error("agent-batch-executor 需要一个输入值，可以是任务描述、项目路径、网站网址或已导出的工件");
+  }
+
+  const format = getFlagValue(args, "--format") || (hasFlag(args, "--json") ? "json" : undefined);
+  const batch = await createAgentBatchExecutor(input, {
+    intent: getFlagValue(args, "--intent"),
+    format,
+    taskId: getFlagValue(args, "--task"),
+    count: getFlagValue(args, "--count")
+  });
+  const outputJson = batch.format === "json";
+  const renderedOutput = outputJson
+    ? `${JSON.stringify(batch, null, 2)}\n`
+    : renderAgentBatchExecutorMarkdown(batch);
+
+  const outputPath = getFlagValue(args, "--out");
+  if (outputPath) {
+    const resolvedOutputPath = await writeAgentBatchExecutorOutput(outputPath, renderedOutput);
+    process.stdout.write(`已保存 agent batch executor：${resolvedOutputPath}\n`);
+    return;
+  }
+
+  process.stdout.write(renderedOutput);
+}
+
 function handleWhere() {
   process.stdout.write(
     [
@@ -761,6 +824,16 @@ export async function runCli(args = []) {
     return;
   }
 
+  if (command === "agent-executor") {
+    await handleAgentExecutor(rest);
+    return;
+  }
+
+  if (command === "agent-batch-executor") {
+    await handleAgentBatchExecutor(rest);
+    return;
+  }
+
   if (command === "skills") {
     await handleSkills(rest);
     return;

package/src/index.js

@@ -7,7 +7,9 @@ export {
 } from "./interactive-onboarding.js";
 export { createAutoFlow, renderAutoFlowMarkdown, writeAutoFlowOutput } from "./auto-flow.js";
 export { createApplyPlan, renderApplyPlanMarkdown, writeApplyPlanOutput } from "./apply-plan.js";
+export { createAgentBatchExecutor, renderAgentBatchExecutorMarkdown, writeAgentBatchExecutorOutput } from "./agent-batch-executor.js";
 export { createAgentHandoff, renderAgentHandoffMarkdown, writeAgentHandoffOutput } from "./agent-handoff.js";
+export { createAgentExecutor, renderAgentExecutorMarkdown, writeAgentExecutorOutput } from "./agent-executor.js";
 export { createAgentRunbook, renderAgentRunbookMarkdown, writeAgentRunbookOutput } from "./agent-runbook.js";
 export { createAgentSession, renderAgentSessionMarkdown, writeAgentSessionOutput } from "./agent-session.js";
 export { createCompletionReport, renderCompletionReportMarkdown, writeCompletionReportOutput } from "./completion-report.js";

package/src/skills.js

@@ -7,6 +7,8 @@ const SKILL_ORDER = [
   "geo-ai-search-optimization-auto-flow",
   "geo-ai-search-optimization-agent-session",
   "geo-ai-search-optimization-agent-runbook",
+  "geo-ai-search-optimization-agent-executor",
+  "geo-ai-search-optimization-agent-batch-executor",
   "geo-ai-search-optimization-usage",
   "geo-ai-search-optimization-agent-handoff",
   "geo-ai-search-optimization-repair-loop",
@@ -23,6 +25,8 @@ const SKILL_CATEGORY = {
   "geo-ai-search-optimization-auto-flow": "routing",
   "geo-ai-search-optimization-agent-session": "routing",
   "geo-ai-search-optimization-agent-runbook": "execution",
+  "geo-ai-search-optimization-agent-executor": "execution",
+  "geo-ai-search-optimization-agent-batch-executor": "execution",
   "geo-ai-search-optimization-usage": "guidance",
   "geo-ai-search-optimization-agent-handoff": "execution",
   "geo-ai-search-optimization-repair-loop": "execution",
@@ -158,6 +162,8 @@ export function renderBundledSkillsMarkdown(bundle) {
     "- 如果 agent 需要自动选 skill，先跑 auto-flow。",
     "- 如果要给 agent 明确步骤，继续进入 agent-session。",
     "- 如果要给 agent 一份执行手册和检查表，再进入 agent-runbook。",
+    "- 如果要直接告诉 agent 这轮先做哪 1 包，再进入 agent-executor。",
+    "- 如果要连续推进前 2 到 3 包，但仍然一次只做一包，再进入 agent-batch-executor。",
     "- 再看 usage skill，知道什么时候该跑哪个命令。",
     "- 如果要交给 agent 执行，再进入 handoff / apply / completion 这一条执行链。",
     "- 如果要产出给团队分发，再进入 share / export / html / publish 这一条交付链。",