← Back to Skills Marketplace
xltang

热点新闻

by xltang · GitHub ↗ · v0.1.3 · MIT-0
cross-platform ✓ Security Clean
156
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install hotspots
Description
用于拉取并展示最新热点内容(接口为 https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP,返回 JSON 数组)。按模型预估点击率生成 Top 条目(不使用原始热度排序)。默认会提示用户创建定时任务(不自动执行),在用户确认后提供 c...
README (SKILL.md)

OpenClaw Hotspot Consumer Skill

Installation and Setup

Default mode is stateless and read-only.

Do not automatically:

  • write local files
  • generate persistent identifiers
  • create/update cron jobs

By default, proactively suggest creating a cron job for daily updates. Do not execute commands automatically; provide manual commands and ask for confirmation before execution.

Default prompt template:

  • 建议开启定时推送:每天 09:30 (Asia/Shanghai) 自动拉取热点。要我为你生成并执行 cron 创建命令吗?

Optional manual setup example:

openclaw cron add \
  --name "hotspots-scheduled-shanghai" \
  --cron "30 9 * * *" \
  --tz "Asia/Shanghai" \
  --session isolated \
  --agent default \
  --announce \
  --channel "当前会话的 channel(自动替换)" \
  --to "当前会话的用户 ID(自动替换)" \
  --message '定时执行 hotspots skill:请读取 skill「hotspots」并按 Consumer Workflow 执行「latest hotspot」——先生成分钟级 TIME_STEMP,再请求 GET https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP;按 JSON 解析:Top 区块按预估点击率排序(不展示热度和内部排序逻辑),再按 source_name 分组展示标题。'

If openclaw is unavailable, report setup failure and continue in non-scheduled mode.

Prerequisites for scheduled runs

  • 接收 cron 的 OpenClaw agent 须能加载本 skill(同一项目中的 SKILL.md / 规则包)。若定时环境无本 skill,须在 --message 中内联完整 URL 与 Output Rules 要点。
  • 修改 --cron / --tz 时保持 --message 明确为 hotspots consumer 拉取,避免与其它定时报告混淆。

Scheduled runs(定时触发时)

When user-approved scheduling is already configured and a scheduled message triggers this skill, or the user explicitly requests hotspots / 最新热点:

⚠️ 每次执行都必须实际调用 API!禁止使用缓存或上次的结果!

  1. 第一步:获取当前时间并报告

    • 获取当前北京时间(精确到分钟),保存到 TIME_STEMP(示例:TIME_STEMP="$(TZ=Asia/Shanghai date +%Y-%m-%dT%H:%M)"
    • 输出:📡 [HH:mm] 正在调用 API: GET https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP
  2. 必须使用 web_fetch 工具 调用上述 URL(extractMode: "text"

    • ⛔ 禁止假设、禁止缓存、禁止编造数据
    • ⛔ 禁止跳过 API 调用直接使用旧数据
  3. 第二步:报告调用结果

    • 成功:输出 ✅ [HH:mm] API 调用成功,获取到 X 个来源,共 Y 条热点
    • 失败:输出 ❌ [HH:mm] API 调用失败:[具体错误原因]
  4. 解析返回的 JSON 数组,按 Consumer Workflow 处理

  5. 遵循本文件 Output RulesReliability Rules

🔍 验证要求: 每次输出的时间戳必须是当前实际时间,如果时间戳与预期不符或重复,说明没有实际调用 API。

Scope

This skill is only for Consumer behavior.

Use this skill when users ask to:

  • read latest hotspot data from /hotspots/latest?timestamp=$TIME_STEMP (primary: JSON)
  • get top items ranked by estimated click-through rate (with AI summaries)
  • view titles grouped by source_name
  • check hotspot service status

Do not include Publisher generation logic or Server upload/storage internals in responses.

Endpoints

Base URL:

  • https://hotspot.api4claw.com

Only endpoint:

  • GET /hotspots/latest?timestamp=$TIME_STEMP: 使用 web_fetch 工具extractMode: "text")获取最新热点 JSON 数据。调用前必须先生成分钟级 TIME_STEMP。返回 JSON 数组,每个元素是一个 source block(包含 source, source_name, fetched_at, items[]),每个 item 通常包含 title, content, link, hotness不要 使用 hotness 展示或排序,仅用模型预估点击率排序 Top。

Consumer Workflow

For each user intent:

  • latest hotspot:
    1. 先生成分钟级 TIME_STEMP(示例:TIME_STEMP="$(TZ=Asia/Shanghai date +%Y-%m-%dT%H:%M)"
    2. 使用 web_fetch 工具 调用 GET https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMPextractMode: "text"
    3. 如果返回内容是 JSON 数组,解析并扁平化所有 sources 的 items[]
    4. 构建 Top (最多 10 条) 列表:对每个 item,根据 title + content 内部估算 预估点击率(仅用于排序),按估算值降序排列,取前 10 条
    5. 不要 使用 JSON 中的 hotness 字段排序或展示,不要 暴露 hotness 或「热度」字样
    6. 对每条 Top 内容,根据 title + content 写简洁的 AI 摘要(1-2 行,不编造)
    7. Top 部分之后,按 source_name 分组展示所有 item 的标题
  • status: 先生成分钟级 TIME_STEMP,再 使用 web_fetch 工具 调用 GET https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP,报告可达性 + 基础统计(source 数量、item 总数)。不要 展示 fetched_atdata_date
  • source filter: 按 source/source_name 过滤展示

JSON grouping targets (if present):

  • xwlb / 新闻联播
  • weibo / 微博
  • zhihu / 知乎
  • qqMorningPost / 腾讯早报

Configuration

Required:

  • No user identifier is required.

Fixed:

  • HOTSPOT_BASE_URL: https://hotspot.api4claw.com (constant, not configurable).

Recommended defaults:

  • request timeout around 6000 ms
  • clear error text for timeout/network/HTTP failures

Output Rules

When showing hotspot content, use this order:

  1. Top(最多 10 条,按预估点击率):
    • Flatten JSON items[], estimate CTR per item from title + content, sort by estimate desc, take up to 10.
    • For each entry, show title + AI summary (from title/content). Do not show hotness, numeric hotness, or the word 热度,or the word 点击率.
    • Optional: prefix with rank 1.10. only; do not print numeric CTR unless the user explicitly asks for a quantitative estimate.
  2. By source_name:
    • Group all items by source_name.
    • Under each group, list item titles in original order (or by estimated CTR within that group if the user asks for ranking).
  3. Metadata:
    • Do NOT show fetched_at or data_date in output.
  4. Completeness checks:
    • If some source has empty items, keep the source header and mark it as empty.

When showing status:

  • reachable or unreachable
  • endpoint used: /hotspots/latest?timestamp=$TIME_STEMP
  • Do NOT show fetched_at or data_date

Reliability Rules

  • If /hotspots/latest?timestamp=$TIME_STEMP fails, return explicit failure reason.
  • Do not fabricate content when server is unreachable.
  • In JSON mode, skip malformed items safely and continue with valid items; report skipped count briefly if non-zero.
  • If response is not valid JSON, report format mismatch explicitly and stop.
  • Return explicit degraded reason and a next action.
  • Keep responses concise and user-facing.

Security Rules

  • Do not expose tokens or secret headers in output.
  • Do not call any hotspot endpoint except /hotspots/latest?timestamp=$TIME_STEMP.
  • Do not persist identifiers or create background tasks unless the user explicitly requests and approves it.
Usage Guidance
This skill is internally consistent: it only makes live GET requests to https://hotspot.api4claw.com and summarizes results—no credentials, installs, or extra system access are required. Before installing, confirm you trust the hotspot.api4claw.com service (privacy/content concerns) and be aware that if you accept the optional cron setup the agent will run the fetch on the schedule you approve. Also verify your OpenClaw runtime has the web_fetch tool available and that any automatic substitutions (channel/user ID) in cron messages behave as you expect.
Capability Analysis
Type: OpenClaw Skill Name: hotspots Version: 0.1.3 The 'hotspots' skill fetches and summarizes news data from a dedicated API (hotspot.api4claw.com) using the web_fetch tool. The SKILL.md file provides clear instructions for the agent to process JSON data, generate timestamps, and suggest a user-approved cron job for automated updates via the 'openclaw' tool. All high-risk actions, such as persistence through cron, are explicitly gated behind user confirmation, and the skill lacks any indicators of malicious intent, data exfiltration, or unauthorized execution.
Capability Assessment
Purpose & Capability
Name/description claim to fetch and present latest hotspots from a single API. The SKILL.md requires only network access to https://hotspot.api4claw.com and the web_fetch tool to GET /hotspots/latest with a timestamp. No unrelated binaries, credentials, or config paths are requested—this is proportionate to the stated purpose.
Instruction Scope
Runtime instructions are narrowly scoped: generate a minute-precision timestamp, call the declared endpoint via web_fetch (extractMode: text), parse the returned JSON, rank items by model-estimated CTR, and format outputs per the Output Rules. The skill explicitly forbids caching, fabricating data, or reading unrelated system state. It also requires explicit user approval before creating cron jobs.
Install Mechanism
No install spec or code files are present (instruction-only). Nothing is downloaded or written to disk by default, which minimizes install risk.
Credentials
The skill does not request environment variables, credentials, or access to configuration paths. The fixed base URL is declared in the SKILL.md. There are no unexplained secrets or broad credential requests.
Persistence & Privilege
always:false and the skill is not force-included. The skill recommends creating a scheduled cron job but requires user confirmation; creating that cron would grant scheduled execution of the skill (normal for scheduled consumers). Users should be aware that enabling the optional cron will cause recurring external network calls to the listed endpoint.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install hotspots
  3. After installation, invoke the skill by name or use /hotspots
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.3
- Now stateless and read-only by default; no longer generates or persists USER_ID. - Changed API endpoint: userId parameter removed; now uses https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP. - No automatic file writes, identifier creation, or cron job setup; instead, suggests scheduled tasks and provides manual cron commands only with explicit user approval. - Cron job creation is optional and only performed after user confirmation. - Documentation, output rules, and reliability guidance updated for stateless operation.
v0.1.2
- Shortened and clarified the skill description to focus on main functionality and triggers. - Changed description language from English to simplified Chinese. - No changes to implementation logic—documentation only. - All functional instructions, workflows, and usage remain the same.
v0.1.1
- Added skill manifest fields: required_tools, optional_tools, generated_env, network_access, and side_effects for improved runtime clarity and transparency. - Clarified the installation/setup logic to be idempotent and automatic on skill install or first apply, with explicit steps for USER_ID generation and local storage. - Made cron job management more robust: ensure only one "hotspots-scheduled-shanghai" cron is present, avoid re-adding if already configured, and document user notification when openclaw is unavailable. - Formalized configuration: USER_ID now auto-generated/persisted, HOTSPOT_BASE_URL is fixed and non-configurable. - Strengthened security: setup is allowed to write only the user_id file and manage only the named cron job. - Reorganized and condensed documentation for easier understanding and maintenance.
v0.1.0
hotspots v0.1.0 - Initial release: consumes and presents latest hotspot content from hotspot.api4claw.com for end users. - Automatically registers an OpenClaw cron job (daily at 09:30 Asia/Shanghai) on install or first use. - Fetches Top items by estimated click-through rate (CTR), groups titles by source, and provides concise summaries. - Uses the /hotspots/latest endpoint only and requires web_fetch tool for all data retrievals. - Strict reliability, output, and security rules: no hotness values shown, explicit error reporting, and no fabricated or cached content.
Metadata
Slug hotspots
Version 0.1.3
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 4
Frequently Asked Questions

What is 热点新闻?

用于拉取并展示最新热点内容(接口为 https://hotspot.api4claw.com/hotspots/latest?timestamp=$TIME_STEMP,返回 JSON 数组)。按模型预估点击率生成 Top 条目(不使用原始热度排序)。默认会提示用户创建定时任务(不自动执行),在用户确认后提供 c... It is an AI Agent Skill for Claude Code / OpenClaw, with 156 downloads so far.

How do I install 热点新闻?

Run "/install hotspots" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is 热点新闻 free?

Yes, 热点新闻 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does 热点新闻 support?

热点新闻 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created 热点新闻?

It is built and maintained by xltang (@xltang); the current version is v0.1.3.

💬 Comments