← 返回 Skills 市场
cpojer

HN Digest

作者 cpojer · GitHub ↗ · v1.0.1
cross-platform ⚠ suspicious
2979
总下载
1
收藏
14
当前安装
2
版本数
在 OpenClaw 中安装
/install hn-digest
功能描述
Fetch and send Hacker News front-page posts on demand. Use when the user asks for HN, says 'hn', 'pull HN', 'hn 10', or specifies a topic like 'hn health', 'hn hacking', or 'hn tech'. Sends N (default 5) posts as individual messages with Title + Link. Exclude crypto.
使用说明 (SKILL.md)

HN Digest

Command format

Interpret a user message that starts with hn as a request for a Hacker News front-page digest.

Supported forms:

  • hn → default 5 posts
  • hn \x3Cn> → n posts
  • hn \x3Ctopic> → filter/boost by topic
  • hn \x3Cn> \x3Ctopic> → both
  • If the user asks for “more” after already seeing some (e.g. “show me top 10–15 since we already did top 10”), treat it as an offset request and use --offset (e.g. offset 10, count 10).

Topics:

  • tech (default)
  • health
  • hacking
  • life / lifehacks

Output requirements

  • Do not send any extra commentary/preamble/epilogue.
  • Send results as individual messages.
  • Each post message must be exactly:
    • first line: the post title
    • second line: \x3Cage> · \x3CcommentCount> comments (age like 45m ago, 6h ago, 3d ago)
    • third line: the Hacker News comments link (https://news.ycombinator.com/item?id=...)
  • After the post messages, send one final message that is the generated image.
    • If the chat provider requires non-empty text for media, use a minimal caption ..
  • Hard exclude crypto.

Procedure

  1. Parse n and topic from the user message.
  2. Fetch + rank items:
    • Run node skills/hn-digest/scripts/hn.mjs --count \x3Cn> --offset \x3Coffset> --topic \x3Ctopic> --format json.
    • Default offset is 0 unless the user explicitly asks for “more/next” after a previous batch.
  3. Send results as N individual messages in the required 3-line format.
  4. Then generate a delightful mood image via Nano Banana, inspired by the posts you just sent:
    • Use skills/hn-digest/scripts/mood_prompt.mjs to build a prompt from the JSON items.
    • Add 3–4 subtle Easter eggs derived from the post themes (no text/logos; keep it fun).
    • Generate and attach the image by running:
      • skills/hn-digest/scripts/generate_mood_nano_banana.sh ./tmp/hn-mood/hn-mood.png \x3Ctopic> \x3Cn> \x3Coffset>
    • Send the generated image as one additional message.

If fetching/ranking fails or returns 0 items:

  • Use https://news.ycombinator.com/ in the browser tool, pick N non-crypto items by judgment, and send them in the same 3-line format.
  • Still generate a mood image (general “HN tech deep dives” vibe) with a banana Easter egg.
安全使用建议
Before installing or enabling this skill, consider the following: (1) The SKILL.md and registry metadata do NOT declare that an image-generation key is required, but the scripts will look for GEMINI_API_KEY (and fall back to reading ~/.openclaw/openclaw.json for nano-banana-pro.apiKey). Decide whether you are comfortable exposing that key or allowing the skill to read your OpenClaw config. (2) The skill will create a local Python virtualenv and pip-install google-genai and pillow — review those packages and be prepared for code to run on your machine. (3) There is an extra image routine (mood_image.py) that would use OPENAI_API_KEY; confirm whether that is intended or dead/leftover code. (4) If you want to proceed, ask the author to: a) explicitly document required env vars and config access in SKILL.md, b) avoid reading ~/.openclaw/openclaw.json (require an explicit env var instead), or c) remove unused credential-using files. If unsure, run the skill in a sandboxed environment or inspect/execute the scripts locally with empty or test credentials first.
功能分析
Type: OpenClaw Skill Name: hn-digest Version: 1.0.1 The skill is classified as suspicious due to the `nano_banana_mood.py` script attempting to load an API key from `~/.openclaw/openclaw.json`. While this is plausibly for its own legitimate use (loading the 'nano-banana-pro' API key), it demonstrates the capability to access sensitive configuration files from the user's home directory, which is a high-risk permission. Other potential prompt injection vectors identified in `SKILL.md` are mitigated by robust argument validation in `hn.mjs` and careful prompt construction in `mood_prompt.mjs`.
能力评估
⚠ Purpose & Capability
The skill claims no required credentials or config, but its included code expects a Gemini API key (GEMINI_API_KEY) and also contains an alternate image routine that would use OPENAI_API_KEY. The skill's purpose (HN digest + mood image) can justify needing an image-generation key, but the SKILL.md/registry metadata do not declare any required env vars or explain where keys must come from.
⚠ Instruction Scope
SKILL.md tells the agent to run bundled node and shell scripts which: call the HN Algolia API (expected), create a local Python venv and pip-install packages, and run a Python image generator. One image generator (nano_banana_mood.py) will attempt to read ~/.openclaw/openclaw.json to extract a nano-banana-pro apiKey. Reading another skill/agent config file is out-of-scope for a simple HN digest and is not declared in the instructions.
ℹ Install Mechanism
No centralized install spec, but the runtime shell script creates ./tmp/hn-venv and pip-installs google-genai and pillow from PyPI. This is traceable (PyPI) and not an arbitrary HTTP download, but it will write files to disk and execute third-party packages — moderate risk and should be disclosed.
⚠ Credentials
The code checks for GEMINI_API_KEY (env) and, if absent, reads ~/.openclaw/openclaw.json to find nano-banana-pro.apiKey. That means the skill can access stored agent/skill config and potentially other skills' API keys. The repository also contains mood_image.py which would use OPENAI_API_KEY (not referenced by the main generate script) — the presence of multiple credential paths with no declaration is disproportionate and surprising.
ℹ Persistence & Privilege
The skill is not force-included (always: false) and does not request to modify other skills. However, it does read ~/.openclaw/openclaw.json, which leaks a higher-privilege config surface to the skill at runtime — this is a privilege elevation of read-access to agent config and should be considered when enabling the skill.
如何使用
  1. 确保已安装 OpenClaw（本地或 Docker 部署）
  2. 在对话框中输入安装命令：/install hn-digest
  3. 安装完成后，直接呼叫该 Skill 的名称或使用 /hn-digest 触发
  4. 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
hn-digest 1.0.1 - No functional or documentation changes; the version was updated without modifications to files or content.
v1.0.0
Fetch and send top Hacker News front-page posts on demand with topic filtering and delightful mood images. - Responds to user requests like "hn", "hn 10", or "hn health" with N (default 5) front-page posts (excluding crypto), each as an individual 3-line message. - Supports topic filters: tech (default), health, hacking, life/lifehacks. - Handles "more" requests using offset to continue from previous results. - After sending posts, generates and sends a mood image inspired by the topics, with playful Easter eggs. - Falls back to manual fetching and a default mood image if automated fetching fails.
元数据
Slug hn-digest
版本 1.0.1
许可证 —
累计安装 14
当前安装数 14
历史版本数 2
常见问题

HN Digest 是什么？

Fetch and send Hacker News front-page posts on demand. Use when the user asks for HN, says 'hn', 'pull HN', 'hn 10', or specifies a topic like 'hn health', 'hn hacking', or 'hn tech'. Sends N (default 5) posts as individual messages with Title + Link. Exclude crypto. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 2979 次。

如何安装 HN Digest？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install hn-digest」即可一键安装，无需额外配置。

HN Digest 是免费的吗？

是的，HN Digest 完全免费（开源免费），可自由下载、安装和使用。

HN Digest 支持哪些平台？

HN Digest 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 HN Digest？

由 cpojer（@cpojer）开发并维护，当前版本 v1.0.1。

推荐 Skills
self-improving agent
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⬇ 463283 · by pskoett
Skill Vetter
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
⬇ 259410 · by spclaudehome
Polymarket
Query Polymarket prediction markets. Check odds, find trending markets, search events, track price movements.
⬇ 232503 · by joelchance
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⬇ 200423 · by ivangdavila
Github
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
⬇ 191115 · by steipete
ontology
Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
⬇ 190156 · by oswalpalash

💬 留言讨论