🌐

Social Content Research Agent Workflow Prompt Guide

5 core tips · 2 copy-ready templates · Updated 2026-05-15

🎯 Core Principles

1. Position the agent as a public content research assistant, not a posting bot

This workflow is for global social content research across Xiaohongshu, TikTok, Reddit, YouTube Shorts, X/Twitter, and Instagram Reels. Its value is not automated posting or engagement. Its value is turning public posts, public comments, and public discussions into audience insights, content angles, and execution plans.

Use cases:
- Extract user pain points from public posts and comments
- Generate content ideas, hooks, scripts, and asset checklists
- Compare angles across platforms
- Turn repeated questions into free tools, templates, newsletters, or SEO pages

Boundaries:
- No bulk engagement, fake engagement, private messaging, or auto-posting
- No bypassing platform restrictions
- No collection of non-public personal data
- Human review is required before publishing

2. Four-step structure: platform → public content → user needs → content opportunities

A safe social research agent should not jump straight into writing. It should first identify the platform and topic, summarize public content, extract user needs, and then generate content opportunities.

Reusable workflow prompt:

You are a social content research agent. Analyze public content only. Do not automate engagement or publishing.
Platform: [Xiaohongshu / TikTok / Reddit / YouTube Shorts / X / Instagram]
Topic: [topic]
Audience: [audience]
Output four sections:
1. What users are searching for: 10 keywords, long-tail terms, and user-language phrases
2. What high-engagement posts have in common: title, opening, structure, comment triggers
3. What comments reveal: pain points, objections, common questions, unmet needs
4. Content opportunities: 10 ideas with platform, title, angle, audience, and risk notesFormat: Markdown table

3. Use Xiaohongshu as one Chinese-market example, not the whole product positioning

Xiaohongshu is useful for Chinese audience insight, but yangmao.ai is a global AI free tools and API intelligence site. So Xiaohongshu should be treated as one platform example alongside TikTok, Reddit, YouTube Shorts, and X/Twitter.

Platform roles:
- Xiaohongshu: Chinese pain points, tutorials, reviews, comparison lists
- TikTok / YouTube Shorts: hooks, short scripts, visual concepts
- Reddit: authentic complaints, alternative searches, long-tail English questions
- X/Twitter: trends, opinions, developer topics, distribution hooks

4. Comment insight is more valuable than headline generation

Headlines alone quickly become generic. The real value is in comments and discussion threads: why people save something, what they do not understand, what templates they ask for, what tools they compare, and where existing solutions fall short.

Comment analysis prompt:

I will provide public comment or discussion summaries.
Extract:
- Top 10 repeated questions
- Real user scenarios
- Needs that can become tutorials, templates, tool pages, or checklists
- Risky claims to avoid
- The next content title, outline, and CTA
- Whether the topic can become a free AI tool or API comparison page

5. Recommended models and free API path

This workflow connects back to the core yangmao.ai value proposition: use free or low-cost models to do useful research work.

Routing:
- Chinese comments / Xiaohongshu: DeepSeek, Kimi, Qwen, Doubao
- English Reddit / YouTube: Gemini, GPT-5.4 Mini, Claude Sonnet
- Batch titles and short scripts: Gemini Flash, DeepSeek V3.2
- Cost-sensitive work: start with free APIs and no-card API options

📋 Copy-ready Prompt Templates

Social content research agent master prompt

You are a social content research agent. Your job is to analyze public posts, public comments, and public discussions, then generate content ideas and an execution plan.

Platform: [Xiaohongshu / TikTok / Reddit / YouTube Shorts / X / Instagram]
Topic: [topic]
Target audience: [audience]
My product/project: [product or project]

Boundaries:
- Analyze only public information and material I provide
- Do not automate engagement, posting, or private messages
- Do not store personal sensitive data
- All output must be reviewed by a human before use

Output:
1. Keyword map: core terms, long-tail terms, user-language phrases
2. User needs: pain points, objections, repeated questions, motivation
3. Content opportunities: 10 ideas
4. Titles: 3 titles per idea
5. Platform fit: short video, image post, long-form post, or tool page
6. Risk notes: claims and actions to avoid
7. Recommended models: prioritize free or low-cost models

Cross-platform comment insight prompt

Here are public comments or discussion summaries about a topic:
[paste public comments or discussion summaries; do not include private personal data]

Analyze:
- What problem are users really trying to solve?
- Which questions repeat enough to deserve a standalone post?
- Which phrases are in the user's own language and can be used in titles?
- Which needs can become a free tool, template, checklist, or API comparison page?
- Which claims are risky and should be avoided?
- Which platform fits best: Xiaohongshu / TikTok / Reddit / YouTube / X?

Give me 10 content ideas scored by search potential, conversion potential, content difficulty, and global adaptability.

← Other Models 🗺️ AI Free Resource Map

💬 Found an error? Help us improve

📱 Telegram ✉️ Email