hackernews-digest

获取 Hacker News 热门文章并生成每日摘要。对每篇文章进行总结、提炼核心要点、给出可实践建议、灵感启发，以及生成社交媒体分享文案（即刻 + 小红书 + Twitter/X）。 触发场景：用户提到 "HN 摘要"、"Hacker News 日报"、"hackernews digest"、"今日 HN"、"...

Security Scan

The skill's goal (build an HN daily digest) is plausible, but the runtime instructions reference undeclared dependencies and perform potentially destructive filesystem actions (deleting files) and automated parallel agent actions without clear safeguards.

Purpose & Capability

The described purpose (fetch HN Top 10, scrape articles, produce summaries and social media copy) matches the instructions. However, the SKILL.md lists dependencies (autocli, WebFetch) yet the skill metadata declares no required binaries or install steps — this mismatch means the agent may assume tooling exists when it is not declared. Also the default BASE_DIR hardcodes a user path (/Users/victor/...) which looks like leftover dev configuration and is out-of-place for a generic skill.

Instruction Scope

Instructions tell the agent to check for and DELETE an existing output file and an entire temp directory for today's digest without requiring user confirmation. They also mandate creating parallel Agents to scrape content and include a fallback to an agent-browser if URLs are inaccessible. These actions expand scope beyond simple retrieval+summary: they modify local filesystem state and spawn autonomous browser agents, which should be explicitly declared and limited.

Install Mechanism

This is an instruction-only skill (no install spec). That reduces risk of arbitrary code installation, but the SKILL.md depends on external tools (autocli, WebFetch, agent-browser) with no install instructions or declared required binaries. The absence of an install mechanism for these tools is an inconsistency that could break runtime behavior or result in unexpected adaptive behavior by the agent to obtain equivalent tooling.

Credentials

The skill requests no credentials or secret environment variables, which is appropriate. It does support an HN_DIGEST_BASE_DIR env var for configuring storage, which is reasonable. The default path, however, is a hardcoded personal path (/Users/victor/...), which is inappropriate for a distributable skill and could lead to accidental writes/deletions if used as-is.

Persistence & Privilege

The skill is not always-enabled and does not request persistent platform privileges. It writes and deletes files under a base directory (its own data), which is within normal behavior for a content-generation skill. The concern is procedural (deleting files without confirmation), not privilege escalation.

What to consider before installing

This skill is coherent in purpose but has several red flags you should consider before installing or invoking it: (1) It expects tools named 'autocli' and 'WebFetch' and may spawn agent-browser instances, yet no install requirements or binaries are declared — confirm those tools are available or add explicit install steps. (2) The default BASE_DIR is a hardcoded personal path (/Users/victor/...), which you should override via the env var or CLI option to a safe directory you control. (3) The runtime deletes existing output and temp directories for today's digest without prompting; ensure you understand and are comfortable with that destructive behavior or require confirmation. (4) Because it spawns parallel agents and uses fallback browsing, consider limiting concurrency and reviewing logs/output to avoid unexpected network activity. Recommended actions: update the skill metadata to declare required binaries and installation instructions, remove or change the hardcoded default path, add a safety confirmation before deleting files, and document the network/browser fallback behavior clearly.

Hacker News 每日摘要技能

获取 Hacker News Top 10 热门文章，并行抓取原文内容，生成摘要、核心要点、可实践建议、灵感启发及社交媒体分享文案，汇总为一篇每日精选文档。

配置

配置方式	优先级	示例
命令行参数 --base-dir	最高	/hackernews-digest --base-dir /tmp/hn
环境变量 HN_DIGEST_BASE_DIR	中	export HN_DIGEST_BASE_DIR=...
默认值	最低	/Users/victor/Desktop/resource/daily-info/hacknews

路径规则（{YYYYMMDD} 为当天日期）：

• 临时目录：{BASE_DIR}/{YYYYMMDD}/temp/
• 最终文档：{BASE_DIR}/{YYYYMMDD}/hackernews-daily-{YYYYMMDD}.md

依赖

• autocli: 获取 Hacker News 热门文章列表
• WebFetch: 抓取文章原文内容

执行流程

Step 1: 初始化

1. 按优先级确定 BASE_DIR（参数 > 环境变量 > 默认值）
2. 获取当天日期，计算 TEMP_DIR 和 OUTPUT_DIR
3. 检查并清理已存在文件：
   • 检查 {OUTPUT_DIR}/hackernews-daily-{YYYYMMDD}.md 是否存在
   • 如果存在，先删除该文件及 {TEMP_DIR} 目录下的所有临时文件 和 {OUTPUT_DIR}/hackernews-daily-{YYYYMMDD}.md
4. mkdir -p {TEMP_DIR}

Step 2: 获取热门文章列表

autocli hackernews top --limit 10 --format json

Step 3: 并行抓取与分析

为每篇文章创建一个并行 Agent，每个 Agent：

1. 使用 WebFetch 抓取原文内容
2. 按模板生成分析内容
3. 写入临时文件 {TEMP_DIR}/{rank}-{sanitized_title}.md

Agent 输出模板：

## [序号]. [文章标题]

> 原文链接：[url]
> 得分：[score] | 评论数：[comments] | 作者：[author]

### 摘要
3-5 句话总结核心内容，准确、信息密度高。

### 核心要点
- （3-5 个关键要点）

### 可实践建议
2-3 条可立即付诸行动的具体建议。

### 灵感启发
思维模型、跨领域启发、值得深入探索的方向。1-2 段。

### 社交媒体分享文案

**即刻：**
口语化、有观点、适度 emoji、200 字以内，含 #HackerNews #技术日报

**小红书：**
标题党风格标题 + 正文，善用 emoji，300 字以内，带话题标签

**Twitter/X：**
使用简体中文, 简洁有力，280 字符以内，带 hashtag

Step 4: 汇总生成最终文档

读取临时文件，按 rank 排序合并，写入 {OUTPUT_DIR}/hackernews-daily-{YYYYMMDD}.md：

---
title: "Hacker News 每日精选 - {YYYY-MM-DD}"
date: {ISO 8601}
description: "今日 Hacker News Top 10 热门文章摘要与深度解读"
tags:
  - HackerNews
  - 技术日报
  - Daily Digest
categories:
  - 技术
---

# Hacker News 每日精选 - {YYYY-MM-DD}

> 本文精选今日 Hacker News 热门 Top 10 文章，为每篇文章提供摘要、核心要点、可实践建议、灵感启发及社交媒体分享文案。

---

{按 rank 顺序合并所有文章分析内容}

---

## 今日总结

（整体观察：共同趋势、值得关注的话题、给读者的建议。3-5 句话。）

注意事项

1. 并行化：10 篇文章必须并行抓取分析
2. 容错：URL 不可访问时, 使用 autocli 动态判断获取 URL 对应内容，如果还是不行使用 agent-browser 获取内容.
3. 语言：中文撰写
4. 文案风格：即刻轻松有态度、小红书吸睛有干货、Twitter 简洁专业