← 返回 Skills 市场

Content Research

Name: Content Research
Author: hazy2go

作者 Hazy · GitHub ↗ · v1.1.0

cross-platform ⚠ suspicious

1807

总下载

当前安装

版本数

在 OpenClaw 中安装

/install content-research

功能描述

Research trending topics and generate platform-specific content. Triggers on "research [topic]", "what's new in [topic]", "content for [platform]", "create p...

使用说明 (SKILL.md)

Content Research

Two-phase workflow: Research → Create

Phase 1: Research

Triggers: research [topic], what's new in [topic]

1. Search

Use web search to find recent news:

web_search(query="[topic] news", freshness="pw")

Query patterns:

News: [topic] news
Reddit: site:reddit.com [topic]
X/Twitter: site:x.com [topic]

2. Fetch Articles

Extract article content:

web_fetch(url="[URL]", maxChars=8000)

3. Filter

7-day cutoff — discard older content
Skip "what is X" explainers
Skip price predictions / TA
Prioritize: launches, partnerships, updates, drama, milestones

4. Present

## [Topic] Research — [Date]

1. **[Headline]** - [Source] - [X days ago]
   [2-3 sentence summary]
   
2. **[Headline]** - [Source] - [X days ago]
   [2-3 sentence summary]

[up to 5 items, newest first]

Phase 2: Content Creation

Triggers: create content for [platform], #3 for reddit

Platform Formats

Hook title (no clickbait)
2-4 conversational paragraphs
Include source link
End with discussion prompt

Angles:

News share — Straightforward reporting
Discussion — "What do you think..."
Analysis — Your take on implications
ELI5 — Simple explanation
Contrarian — Devil's advocate

X/Twitter

Under 280 chars (or thread)
Hook first line
Line breaks for readability

Angles:

Breaking — Just facts, urgency
Hot take — Engagement bait opinion
Thread — Multi-tweet breakdown
Quote dunk — React to announcement
Meme — Casual/funny

Discord

Bullet lists (no tables)
Wrap links: \x3Chttps://...>
Bold/CAPS for emphasis

Angles:

Alert — One-liner + link
Summary — Key bullets
Discussion — Ask for reactions
Thread — Detailed breakdown
Meme — Community vibe

Professional tone
Lead with insight
3-5 short paragraphs
End with question

Angles:

Industry insight — What it means
Lessons — What we learn
Prediction — Where it's heading
Career — Professional implications
Case study — Deep dive

Brand Voice (Optional)

For branded content, create a brand-config.md file with your voice guidelines:

# Brand: [Name]

## Voice
- [Tone descriptor]
- [Communication style]

## Avoid
- [Things not to say]

## Include
- [Required elements]

When generating branded content, reference your brand config for consistency.

Example Session

User: research defi

Agent: [Returns 5 findings from past 7 days]

User: 2 for reddit

Agent: [5 Reddit angles for finding #2]

User: angle 3

Agent: [Ready-to-post content]

安全使用建议

This skill appears to do exactly what it says: search the web for recent items and generate platform-formatted posts. Before installing or using it, consider: (1) the agent will fetch arbitrary URLs — make sure your environment's web_fetch is appropriately network-restricted (to avoid internal/SSRF exposure); (2) do not paste API keys, passwords, or other secrets into brand-config.md or prompts; (3) review generated content for accuracy and platform policy compliance (the README/example mention 'Karma accounts' which could imply tactics that violate site rules); (4) if you choose to supply a Brave Search API key, only provide a minimal-scope key and treat it as optional. If you want deeper assurance, provide the platform's web_fetch/web_search implementation details or confirm how network access is sandboxed.

功能分析

Type: OpenClaw Skill Name: content-research Version: 1.1.0 The skill utilizes `web_search` and `web_fetch` tools as defined in `SKILL.md`, which grant network access and information retrieval capabilities. User-controlled input (`[topic]`) is directly templated into the `web_search` query. While the skill's instructions do not exhibit explicit malicious intent, this direct templating creates a potential prompt injection vulnerability against the OpenClaw agent if user input is not adequately sanitized before being passed to these powerful tools. This represents a risky capability without clear malicious intent, aligning with a 'suspicious' classification due to the inherent attack surface.

能力评估

✓ Purpose & Capability

Name/description (research trending topics, generate platform-specific content) matches the SKILL.md and README instructions. The skill is instruction-only and does not declare unrelated binaries, config paths, or credentials. The README's note about an optional Brave Search API key is consistent with an optional enhancement and is not required.

ℹ Instruction Scope

Runtime instructions are limited to web_search and web_fetch calls, filtering, summarization, and content formatting per platform — all within the stated purpose. Caution: web_fetch/web_search will retrieve arbitrary web content based on queries (including URLs the user supplies), and the examples/brand-config mention items like 'Karma accounts' which implies guidance about account usage that could encourage gaming/platform policy violations. The skill does not instruct reading unrelated local files or environment variables, but users should avoid putting secrets into brand-config.md or prompts.

✓ Install Mechanism

No install spec and no code files — instruction-only skill. This is low-risk from an install/execution standpoint because nothing is written to disk or downloaded by the skill itself.

✓ Credentials

No required environment variables or credentials are declared. The README's optional Brave Search API key is plausible for search integration; it is optional and not requested by the skill at install time. Users should not paste unrelated secrets into example config files.

✓ Persistence & Privilege

always is false and disable-model-invocation is false (normal). The skill does not request permanent presence or attempt to modify other skills or agent-wide settings.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install content-research
安装完成后，直接呼叫该 Skill 的名称或使用 /content-research 触发
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

v1.1.0

Simplified for cleaner security profile: removed cron scheduling, memory deduplication, browser profiles. Stateless and focused.

v1.0.0

Initial release

元数据

Slug content-research

版本 1.1.0

许可证 —

累计安装 22

当前安装数 20

历史版本数 2

常见问题