agentcrew-mcn 0.1.0__py3-none-any.whl

agentcrew_mcn-0.1.0.dist-info/METADATA

@@ -0,0 +1,160 @@
+Metadata-Version: 2.4
+Name: agentcrew-mcn
+Version: 0.1.0
+Summary: AI MCN 自动推广工具 — 你的AI营销团队，24小时在线工作，不领工资。
+Author: super-rick
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/super-rick/agent-crew
+Project-URL: Bug Reports, https://github.com/super-rick/agent-crew/issues
+Project-URL: Source, https://github.com/super-rick/agent-crew
+Keywords: ai,mcn,marketing,agent,automation,content-marketing,multi-agent
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Internet :: WWW/HTTP :: Dynamic Content
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: openai>=1.0.0
+Requires-Dist: httpx>=0.27.0
+Requires-Dist: click>=8.1.0
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: rich>=13.0.0
+Requires-Dist: chromadb>=0.5.0
+Requires-Dist: schedule>=1.2.0
+Requires-Dist: duckduckgo-search>=6.0.0
+Requires-Dist: streamlit>=1.28.0
+Requires-Dist: pandas>=2.0.0
+Requires-Dist: plotly>=5.0.0
+Dynamic: license-file
+
+# AgentCrew 🤖
+
+<div align="center">
+
+**你的 AI 营销团队，24 小时在线工作，不领工资。**
+
+[![Python Version](https://img.shields.io/badge/python-3.10%2B-blue)]()
+[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)]()
+
+</div>
+
+---
+
+## 什么是 AgentCrew？
+
+AgentCrew 是一个开源的多 Agent 内容营销自动化工具。它由一组 AI"员工"组成，自动完成从内容生成到跨平台分发的全流程。
+
+- **Writer Agent** — 文案员工：生成技术文章、帖子、Thread
+- **Publisher Agent** — 运营员工：发布到掘金、知乎、X/Twitter
+- **Analyst Agent** — 数据分析员工：追踪效果、优化策略（v2）
+
+## 核心特性
+
+- 🧠 **多 Agent 架构** — 每个"员工"独立部署，可插拔
+- 🔧 **Skills + Tools 系统** — 原子化操作 + 高级能力编排
+- 📚 **RAG 知识库** — 基于历史内容和竞品情报的增强生成
+- 🎯 **跨平台发布** — 支持掘金、知乎、X/Twitter 等平台
+- ⏰ **智能调度** — 带随机抖动的定时发布，避免平台检测
+- 🔮 **MCP 预留** — 架构预留 MCP 协议接口（v2 路线图）
+- 📊 **Dashboard** — Web 面板查看发布数据和 Agent 状态（v0.3）
+- 🔁 **自推广 Meta 闭环** — 项目自己推广自己
+
+## 快速开始
+
+### 安装
+
+```bash
+# 1. 克隆项目
+git clone https://github.com/super-rick/agent-crew.git
+cd agent-crew
+
+# 2. 安装依赖
+pip install -r requirements.txt
+
+# 3. 配置环境变量
+cp .env.example .env
+# 编辑 .env，填入你的 API Key
+```
+
+### 配置
+
+编辑 `config.yaml`，填入各平台的认证信息。
+
+```yaml
+# config.yaml 核心配置
+llm:
+  api_key: ${DEEPSEEK_API_KEY}  # 从 .env 读取
+  model: deepseek-chat
+
+platforms:
+  juejin:
+    cookie: ${JUEJIN_COOKIE}    # 浏览器登录后导出 Cookie
+```
+
+### 使用
+
+```bash
+# 生成一篇技术文章
+agent-crew write generate --topic "Python异步编程" --style technical
+
+# 发布到掘金
+agent-crew publish post --file article.md --platform juejin
+
+# 启动定时发布（每6小时）
+agent-crew schedule start --topic-file topics.txt --platform juejin --interval 6
+
+# 管理 RAG 知识库
+agent-crew rag ingest --file article.md --source "my_blog"
+agent-crew rag search --query "AI Agent 架构"
+```
+
+## 架构概览
+
+```
+┌─────────────────────────────────────────────────────┐
+│                   Orchestrator                       │
+│              任务分派 / 调度引擎 / 配置管理              │
+└──┬──────────┬──────────────┬────────────────────────┘
+   │          │              │
+┌──▼──────┐ ┌▼────────┐ ┌──▼────────┐
+│ Writer  │ │Publisher│ │  Analyst  │
+│ Agent   │ │Agent    │ │  Agent    │
+│         │ │         │ │           │
+│ Skills: │ │ Skills: │ │ Skills:   │
+│ 追热点   │ │ 定时分发 │ │ 效果分析   │
+│ 技术文章 │ │ 多平台  │ │ 趋势报告   │
+│ Thread   │ │ 批量    │ │           │
+│         │ │         │ │           │
+│ Tools:  │ │ Tools:  │ │ Tools:    │
+│ search  │ │ twitter │ │ analytics │
+│ rag     │ │ juejin  │ │ compare   │
+│ compose │ │ zhihu   │ │ report    │
+└─────────┘ └─────────┘ └───────────┘
+```
+
+## 技术栈
+
+| 组件 | 技术 |
+|------|------|
+| 语言 | Python 3.10+ |
+| CLI | Click + Rich |
+| LLM | DeepSeek API (OpenAI 兼容) |
+| 向量库 | ChromaDB |
+| 自动化 | Playwright |
+| Dashboard | Streamlit + Plotly (v0.3) |
+
+## 开发路线图
+
+- **v0.1** ✅ Writer Agent + Publisher Agent + CLI + 掘金/知乎
+- **v0.2** 📝 RAG 深度集成 + X/Twitter + 内容策略引擎
+- **v0.3** 📊 Streamlit Dashboard + Analyst Agent
+- **v1.0** 🎯 MCP 协议 + Skill Store + 社区插件
+
+## License
+
+MIT

agentcrew_mcn-0.1.0.dist-info/RECORD

@@ -0,0 +1,41 @@
+agentcrew_mcn-0.1.0.dist-info/licenses/LICENSE,sha256=pWiDBWmgHOAHWHO7oWN9wrGIk5uHsR62OiDffMNGDhM,1068
+agents/__init__.py,sha256=dW97M43ZI875UMaVpwFrGleTaIE99hCbmLBbXN7Zdx0,120
+agents/analyst.py,sha256=GksoGdySl0UTZeryrn6u8unjbT2ofNmrZ_fA5WuOmE0,14869
+agents/base.py,sha256=AMDS6etwq_KBX0xuz2_g-hnIw-wVto7F4J58DUAqVns,3695
+agents/publisher.py,sha256=8vCl_mIoVfboJDOAzaApn_Eg1fcxxPaY2IT-hTP3up0,9098
+agents/skills.py,sha256=7W1D2W_ZSHkNK_hc9bfZ8CvMgaVfzd0LbeveYod6pFE,7091
+agents/tools.py,sha256=ZxgyOaIlvS-VXVQnDOBTt0sufjm4KBiPbEm04HzvArA,5970
+agents/writer.py,sha256=L9O_VBOjX1bopnFvY1eQrW5qfDj-TTbkW8HaVYXRYWQ,10750
+cli/__init__.py,sha256=hNiA5CRHyJsQDmBH6nTxPsGi8OVAvpoZBR41AE_iMMI,43
+cli/analyst.py,sha256=Fa74cVLIqFH96LotMjc4dHJ9s0K3Yg843-izu5j_7RY,7292
+cli/main.py,sha256=KVWgVizq0OmXabfzTvIWj6AeMk21vX7W0sXxE2iI2zM,6533
+cli/publish.py,sha256=-IiueIBvp5zteYQ_FbzU5YX1dYNizarBhQ-0iny_WrA,6513
+cli/rag_cmd.py,sha256=5qqngrJ2y2mNjRd5KXJCS_0BrfYrFtvnGbfautNcivQ,6242
+cli/schedule.py,sha256=Cslv2uLL_a6SPDw0UypGd0XOv6iZUxz_zOslb08gp7o,5950
+cli/write.py,sha256=1Ekfgsh3mkL4OjH1aY2TZ9PV_NurRIYtu1qGOdFhrPY,5486
+dashboard/__init__.py,sha256=IhhDlwJtlXQ1btV91XY9UCKQ9W8-lMHjGc20AuEdO0c,60
+dashboard/app.py,sha256=Hkhbr-xWDBoT80MumcWt0CzpsrmAMRM_gQ6TqjKsJqk,5412
+dashboard/components.py,sha256=Kcgp7lhT58R4sbpCXSW3Lo9ZD0gH9BMShmBBHu9eTTU,7199
+dashboard/data_loader.py,sha256=ECB9-eKXBWYtUve8Jdvc8GF3dUlvxUi5Ajbe6MT7bm0,6022
+dashboard/pages/__init__.py,sha256=_Q80Xal7cQFs9RD1Rp7CteTNr4Pxzg4Ij2Onx2Lx5Cs,31
+dashboard/pages/analytics.py,sha256=HXJqeU8KEBwkvgBecyOKhlf5YfZS-qT71oh-U2E9Yeo,4383
+dashboard/pages/publishing.py,sha256=6Zv55nqoP9yJTmgjfXAP1fl0489JlJnmGxGfwV1sK5Q,3169
+dashboard/pages/system.py,sha256=el_MQH2spmqtCVWp21IbSQ-wJdwP9YOVq1rQ0XvvE8k,3263
+llm/__init__.py,sha256=nH01TcT4p0OsXfEcwvTkAJuWDALrn1_fs4NGcFhkg5k,129
+llm/client.py,sha256=a-ltaD0-y_X8cGDsARHSbGE5y6lH1Zri2lAq_JFwI3A,2803
+orchestrator/__init__.py,sha256=SeyxV-nEBS0prfA0V4jVnOykfxoQa21L87PeHG_BNTw,52
+orchestrator/manager.py,sha256=n7l6I397__6h_Z7QUEXBZzWsbgzX2l6BR6WGI2DuhEI,8933
+orchestrator/scheduler.py,sha256=pcNaGBFBZ6tPaP8wVSIdjww6adm4LRx-hr-o2cWZh5M,5902
+platforms/__init__.py,sha256=iF3yrPVPUFgmLnYwyCc-XgtiAXeSJhLBCaCXHP02jd8,57
+platforms/base.py,sha256=-iFQOQ99NHCq_LGSaSUAlj0z9OrcdPMCoXhFygqk_Hc,3578
+platforms/juejin.py,sha256=bdscTIVVKmvudcPfkOdOyZRjrwrn-1Q3Jl25grio8uo,7716
+platforms/zhihu.py,sha256=cDL7JDxNmfWKPpsp3netNxDIKCYXnJB-GWq3GKYTMQc,9431
+rag/__init__.py,sha256=5SWzdS-WPJTmrl-ocHlwg1gsbs2ffboNqZu7HVdQlpA,43
+rag/embedder.py,sha256=9Ybw8PCLzJqsEiuS4v6-Y8HbaOpSLYCfBeaTQCYdT6w,1287
+rag/knowledge_base.py,sha256=_0-OPrzKpckJTnIeM-1I4geKySxArB_IW7oMX3_NKm0,4822
+rag/retriever.py,sha256=QqtK3CIoIPy_5HEYAS-S76v-_CMsQ4oookQMnrwZEm8,2027
+agentcrew_mcn-0.1.0.dist-info/METADATA,sha256=Je7izE0alkipUD_3nzyfvAEKhZ-xAgVoYkdxVzpGcoA,5525
+agentcrew_mcn-0.1.0.dist-info/WHEEL,sha256=K260EYznzXsJYBQGqmI8VTxEdiZYNvDZwW9cBh9-_MA,91
+agentcrew_mcn-0.1.0.dist-info/entry_points.txt,sha256=mtl8Gy_9FaMkpkJMhWUL-RpBXFhNX2rmgMA9KaN9b0I,45
+agentcrew_mcn-0.1.0.dist-info/top_level.txt,sha256=ORi6zM-akZpk2mzc-czgFr7DJj7yvpErn_tscUvzGyY,52
+agentcrew_mcn-0.1.0.dist-info/RECORD,,

agentcrew_mcn-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (83.0.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

agentcrew_mcn-0.1.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+agent-crew = cli.main:main

agentcrew_mcn-0.1.0.dist-info/licenses/LICENSE

@@ -0,0 +1,19 @@
+Copyright (c) {{ year }} {{ organization }}
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
+DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE
+OR OTHER DEALINGS IN THE SOFTWARE.

agentcrew_mcn-0.1.0.dist-info/top_level.txt

@@ -0,0 +1,7 @@
+agents
+cli
+dashboard
+llm
+orchestrator
+platforms
+rag

agents/__init__.py

@@ -0,0 +1,5 @@
+"""Agent module."""
+
+from agents.base import BaseAgent, Task, TaskResult
+
+__all__ = ["BaseAgent", "Task", "TaskResult"]

agents/analyst.py

@@ -0,0 +1,370 @@
+from __future__ import annotations
+"""
+Analyst Agent — AI 数据分析员工。
+
+负责：
+1. 效果分析（analyze）— 聚合发布历史数据，统计各平台表现
+2. 报告生成（report）— 调用 LLM 生成可读的效果报告
+3. 策略建议（recommend）— 基于数据生成内容策略优化建议
+
+数据源：直接读取 data/post_history.json（独立于 PublisherAgent）。
+核心流程：
+    execute(task) →
+      task.params.action: "analyze" | "report" | "recommend"
+      → _load_history() → _filter → _calculate_metrics → (LLM 可选) → TaskResult
+"""
+
+import json
+from collections import defaultdict
+from datetime import datetime, timedelta
+from pathlib import Path
+from typing import Any
+
+from agents.base import BaseAgent, Task, TaskResult
+from llm.client import LLMClient
+
+
+class AnalystAgent(BaseAgent):
+    """Content performance analysis agent — the AI data analyst employee.
+
+    Independent of PublisherAgent — reads post history directly from the
+    shared history file (data/post_history.json).
+    """
+
+    name = "analyst"
+    description = "负责内容效果数据分析、报告生成和策略优化建议"
+
+    def __init__(
+        self,
+        llm_client: LLMClient,
+        config: dict | None = None,
+    ):
+        super().__init__(llm_client, config)
+        self._history_file: str = (
+            config.get("history_file", "data/post_history.json")
+            if config
+            else "data/post_history.json"
+        )
+        self._default_days: int = (
+            config.get("default_days", 7) if config else 7
+        )
+
+    def get_system_prompt(self) -> str:
+        return (
+            "你是一个专业的内容运营数据分析师。"
+            "你通过分析各平台发布数据，识别内容表现趋势和优化机会。"
+            "你的分析结论应当数据驱动、可执行、有具体行动建议。"
+            "你使用中文输出，语言简洁专业。"
+        )
+
+    # ── Public API ──────────────────────────────────────────
+
+    def execute(self, task: Task) -> TaskResult:
+        """Execute an analyst task.
+
+        Task params:
+            action (str): analyze | report | recommend (default: analyze)
+            days (int): 分析最近多少天的数据，默认取 config.default_days (7)
+            platforms (list[str], optional): 按平台过滤
+        """
+        started_at = datetime.now()
+        action = task.params.get("action", "analyze")
+        days = task.params.get("days", self._default_days)
+        platforms = task.params.get("platforms")
+
+        try:
+            # Step 1: Collect and filter data
+            history = self._load_history()
+            filtered = self._filter_by_days(history, days)
+            if platforms:
+                filtered = [r for r in filtered if r.get("platform") in platforms]
+
+            # Step 2: Calculate metrics
+            analysis = self._calculate_metrics(filtered, days)
+
+            if action == "analyze":
+                completed_at = datetime.now()
+                return TaskResult(
+                    task_id=task.task_id,
+                    success=True,
+                    data={
+                        "action": "analyze",
+                        "period_days": days,
+                        **analysis,
+                    },
+                    started_at=started_at,
+                    completed_at=completed_at,
+                    duration_seconds=(completed_at - started_at).total_seconds(),
+                    agent_name=self.name,
+                )
+
+            elif action == "report":
+                report_text = self._generate_report(analysis)
+                completed_at = datetime.now()
+                return TaskResult(
+                    task_id=task.task_id,
+                    success=True,
+                    data={
+                        "action": "report",
+                        "period_days": days,
+                        "report": report_text,
+                        **analysis,
+                    },
+                    started_at=started_at,
+                    completed_at=completed_at,
+                    duration_seconds=(completed_at - started_at).total_seconds(),
+                    agent_name=self.name,
+                )
+
+            elif action == "recommend":
+                recommend_text = self._generate_recommendations(analysis)
+                completed_at = datetime.now()
+                return TaskResult(
+                    task_id=task.task_id,
+                    success=True,
+                    data={
+                        "action": "recommend",
+                        "period_days": days,
+                        "recommendations": recommend_text,
+                        **analysis,
+                    },
+                    started_at=started_at,
+                    completed_at=completed_at,
+                    duration_seconds=(completed_at - started_at).total_seconds(),
+                    agent_name=self.name,
+                )
+
+            else:
+                raise ValueError(
+                    f"Unknown action for analyst: {action!r}. "
+                    f"Expected: analyze, report, recommend"
+                )
+
+        except Exception as e:
+            completed_at = datetime.now()
+            return TaskResult(
+                task_id=task.task_id,
+                success=False,
+                error_message=str(e),
+                started_at=started_at,
+                completed_at=completed_at,
+                duration_seconds=(completed_at - started_at).total_seconds(),
+                agent_name=self.name,
+            )
+
+    # ── Data Loading ────────────────────────────────────────
+
+    def _load_history(self) -> list[dict]:
+        """Load post history from the shared history file.
+
+        Independent of PublisherAgent — reads directly from the JSON file
+        that PublisherAgent writes to. Returns empty list if the file
+        doesn't exist or is malformed (first run, no posts yet).
+        """
+        path = Path(self._history_file)
+        if not path.exists():
+            return []
+        try:
+            with open(path, "r", encoding="utf-8") as f:
+                return json.load(f)
+        except (json.JSONDecodeError, OSError):
+            return []
+
+    @staticmethod
+    def _filter_by_days(records: list[dict], days: int) -> list[dict]:
+        """Keep only records from the last N days."""
+        if not records:
+            return []
+        cutoff = datetime.now() - timedelta(days=days)
+        result: list[dict] = []
+        for r in records:
+            posted_str = r.get("posted_at")
+            if not posted_str:
+                continue
+            try:
+                posted_dt = datetime.fromisoformat(posted_str)
+                if posted_dt >= cutoff:
+                    result.append(r)
+            except (ValueError, TypeError):
+                continue
+        return result
+
+    # ── Metrics Calculation ─────────────────────────────────
+
+    def _calculate_metrics(self, records: list[dict], days: int) -> dict[str, Any]:
+        """Aggregate post records into structured metrics.
+
+        Returns dict with:
+            total_posts, success_count, fail_count, success_rate
+            platform_stats: [{platform, total, success, fail, rate}, ...]
+            daily_counts: [{date, total, success, fail}, ...]
+            recent_fails: [{platform, error, posted_at}, ...]
+            error_summary: {error_type: count, ...}
+        """
+        total = len(records)
+        success = sum(1 for r in records if r.get("success"))
+        failed = total - success
+        success_rate = (success / total * 100) if total else 0.0
+
+        # Per-platform stats
+        platform_buckets: dict[str, dict] = defaultdict(
+            lambda: {"total": 0, "success": 0, "fail": 0}
+        )
+        for r in records:
+            plat = r.get("platform", "unknown")
+            platform_buckets[plat]["total"] += 1
+            if r.get("success"):
+                platform_buckets[plat]["success"] += 1
+            else:
+                platform_buckets[plat]["fail"] += 1
+
+        platform_stats: list[dict] = []
+        for plat, pb in sorted(platform_buckets.items()):
+            p_rate = (pb["success"] / pb["total"] * 100) if pb["total"] else 0.0
+            platform_stats.append({
+                "platform": plat,
+                "total": pb["total"],
+                "success": pb["success"],
+                "fail": pb["fail"],
+                "success_rate": round(p_rate, 1),
+            })
+
+        # Daily counts (fill gaps with zero)
+        now = datetime.now()
+        daily_map: dict[str, dict] = defaultdict(
+            lambda: {"total": 0, "success": 0, "fail": 0}
+        )
+        for r in records:
+            posted_str = r.get("posted_at")
+            if not posted_str:
+                continue
+            try:
+                d = datetime.fromisoformat(posted_str).strftime("%Y-%m-%d")
+                daily_map[d]["total"] += 1
+                if r.get("success"):
+                    daily_map[d]["success"] += 1
+                else:
+                    daily_map[d]["fail"] += 1
+            except (ValueError, TypeError):
+                continue
+
+        daily_counts: list[dict] = []
+        for i in range(days - 1, -1, -1):
+            day_label = (now - timedelta(days=i)).strftime("%Y-%m-%d")
+            d = daily_map.get(day_label, {"total": 0, "success": 0, "fail": 0})
+            daily_counts.append({
+                "date": day_label,
+                "total": d["total"],
+                "success": d["success"],
+                "fail": d["fail"],
+            })
+
+        # Recent failures (up to 5)
+        recent_fails = [
+            {
+                "platform": r.get("platform"),
+                "error": r.get("error_message"),
+                "posted_at": r.get("posted_at"),
+            }
+            for r in records
+            if not r.get("success")
+        ][-5:]
+
+        # Error type summary
+        error_counts: dict[str, int] = defaultdict(int)
+        for r in records:
+            if not r.get("success") and r.get("error_message"):
+                err = r["error_message"]
+                key = err.split(":")[0].strip() if ":" in err else err[:80]
+                error_counts[key] += 1
+
+        return {
+            "total_posts": total,
+            "success_count": success,
+            "fail_count": failed,
+            "success_rate": round(success_rate, 1),
+            "platform_stats": platform_stats,
+            "daily_counts": daily_counts,
+            "recent_fails": recent_fails,
+            "error_summary": dict(
+                sorted(error_counts.items(), key=lambda x: -x[1])
+            ),
+        }
+
+    # ── LLM-powered Generation ──────────────────────────────
+
+    def _generate_report(self, metrics: dict[str, Any]) -> str:
+        """Generate a human-readable performance report via LLM."""
+        prompt = self._build_report_prompt(metrics)
+        messages = self._build_messages(prompt)
+        return self.llm_client.chat(messages)
+
+    def _generate_recommendations(self, metrics: dict[str, Any]) -> str:
+        """Generate strategy optimisation recommendations via LLM."""
+        prompt = self._build_recommend_prompt(metrics)
+        messages = self._build_messages(prompt)
+        return self.llm_client.chat(messages)
+
+    # ── Prompt Builders ─────────────────────────────────────
+
+    def _build_report_prompt(self, metrics: dict) -> str:
+        """Build the LLM prompt for report generation."""
+        plat_rows = "\n".join(
+            f"  - {p['platform']}: 共发布 {p['total']} 篇, "
+            f"成功 {p['success']} 篇, 失败 {p['fail']} 篇, "
+            f"成功率 {p['success_rate']}%"
+            for p in (metrics.get("platform_stats") or [])
+        )
+
+        daily_rows = "\n".join(
+            f"  - {d['date']}: 发布 {d['total']} 篇 "
+            f"(成功 {d['success']}, 失败 {d['fail']})"
+            for d in (metrics.get("daily_counts") or [])
+        )
+
+        fail_rows = "\n".join(
+            f"  - [{f.get('platform', '?')}] {f.get('error', '未知错误')}"
+            for f in (metrics.get("recent_fails") or [])
+        )
+
+        return (
+            "请根据以下发布数据生成一份内容运营周报。\n\n"
+            f"## 总览\n"
+            f"- 统计周期: 过去 {metrics.get('period_days', '?')} 天\n"
+            f"- 发布总数: {metrics['total_posts']} 篇\n"
+            f"- 成功: {metrics['success_count']} 篇 ({metrics['success_rate']}%)\n"
+            f"- 失败: {metrics['fail_count']} 篇\n\n"
+            f"## 各平台表现\n{plat_rows if plat_rows else '  (无数据)'}\n\n"
+            f"## 每日发布\n{daily_rows if daily_rows else '  (无数据)'}\n\n"
+            f"## 近期失败\n{fail_rows if fail_rows else '  (无失败记录)'}\n\n"
+            "请生成周报，包含：\n"
+            "1. 核心发现（3-5 条关键结论）\n"
+            "2. 各平台表现评价\n"
+            "3. 失败原因分析与改进建议\n"
+            "4. 下一步行动计划\n\n"
+            "格式：Markdown，标题层级分明，数据引用准确。"
+        )
+
+    def _build_recommend_prompt(self, metrics: dict) -> str:
+        """Build the LLM prompt for strategy recommendations."""
+        plat_rows = "\n".join(
+            f"  - {p['platform']}: {p['total']} 篇, "
+            f"成功率 {p['success_rate']}%"
+            for p in (metrics.get("platform_stats") or [])
+        )
+
+        return (
+            "你是一个内容运营策略顾问。请根据以下发布数据给出优化建议。\n\n"
+            f"## 当前数据\n"
+            f"- 统计周期: 过去 {metrics.get('period_days', '?')} 天\n"
+            f"- 发布总数: {metrics['total_posts']} 篇\n"
+            f"- 成功率: {metrics['success_rate']}%\n\n"
+            f"## 各平台\n{plat_rows if plat_rows else '  (无数据)'}\n\n"
+            "请从以下维度给出具体可执行的策略建议：\n"
+            "1. 平台策略 — 哪些平台应加大/减少投入\n"
+            "2. 发布频率 — 建议的节奏\n"
+            "3. 内容方向 — 什么内容类型效果更好\n"
+            "4. 发布时机 — 最佳发布时间建议\n"
+            "5. 技术改进 — 降低发布失败率的措施\n\n"
+            "建议应数据驱动、具体可执行，给出优先级排序。"
+        )