Community Demand Prospecting
Overview
Turn vague growth ideas into a repeatable workflow: find relevant community conversations, estimate intent, avoid bad-fit threads, and draft replies that help first and promote only when appropriate.
Default to human-in-the-loop. This skill is for demand capture and community prospecting, not mass posting or deceptive automation.
It is especially useful for repo audit, launch readiness, market research, competitor research, positioning, finding users for a product, and planning Reddit or X outreach.
Quick Start
Collect a compact product brief before doing outreach research. Ask for only what is missing:
- - Product name and one-line description
- Target user or buyer
- Main pain points solved
- Best differentiators
- Allowed claims and forbidden claims
- Desired tone
- Destination link, demo, or CTA
If the user has not provided a brief, use the template in assets/product-brief-template.md.
If the user only provides a repository, landing page, or README, start with a launch-readiness audit and market scan before doing community prospecting. Use:
- - INLINECODE1
- INLINECODE2
- INLINECODE3
- INLINECODE4
- INLINECODE5
- INLINECODE6
Operating Sequence
Run the skill in this order unless the user explicitly asks for a narrower deliverable:
- 1. Extract or complete the product brief.
- Audit launch readiness if the input is a repo, README, or rough landing page.
- Research users, competitors, substitutes, and category shape.
- Choose one primary positioning angle.
- Translate the product into user pain-language.
- Generate search queries and locate candidate threads.
- Score each thread for intent and risk.
- Choose the engagement ladder.
- Draft outreach in platform-native style.
- Return a decision-oriented report.
Concrete Example
Example request:
CODEBLOCK0
Example output shape:
CODEBLOCK1
Workflow
Follow this sequence unless the user asks for only one narrow output.
1. Build the Product Brief
If the user gives a repo instead of a polished product brief, derive the brief from:
- - README headline and subhead
- install steps
- release assets
- screenshots or demo assets
- stated use cases
- obvious trust blockers such as unsigned binaries or missing docs
Call out launch blockers before recommending outreach. Common blockers:
- - no screenshot or demo
- weak headline
- unclear audience
- missing install path
- trust friction
- no clear CTA
- overly narrow positioning
If needed, read references/repo-launch-readiness.md.
2. Research the Market and User Landscape
Before drafting outreach, understand the category:
- - who the likely user is
- what job they are hiring the product for
- what alternatives or substitutes they already mention
- whether the market is crowded, fragmented, or lightly served
- what angle is most differentiated: speed, simplicity, price, trust, control, niche fit, or workflow fit
When researching similar products, capture:
- - direct competitors
- indirect substitutes
- incumbent tools
- manual workaround patterns
When researching users, capture:
- - where they ask for help
- how they describe the pain
- what triggers them to switch
- what objections or trust concerns appear repeatedly
Do not reduce market research to a feature checklist. Prefer:
- - positioning differences
- target user differences
- trust and adoption friction
- discovery channel differences
If needed, read references/market-research.md and use assets/market-scan-template.md.
3. Choose the Positioning Angle
After the market scan, decide what the product should lead with.
Choose the main angle from evidence, not preference:
- - problem: the pain is obvious and urgent
- audience: the product is clearly for a narrow group with shared identity
- comparison: users already compare products in this category
- simplicity: incumbents feel bloated, complex, or expensive
- trust: users fear failure, security risk, or setup friction
- workflow fit: the product fits a specific repeated job better than generic tools
Pick one primary angle and at most one secondary angle.
Avoid stacking too many angles into the same message. A weak “everything tool” message usually performs worse than a narrow but clear one.
If needed, read references/positioning-heuristics.md and use assets/positioning-brief-template.md.
4. Build the Demand Map
Translate the product brief into how real people describe the problem:
- - Frustrations
- Current workaround
- Desired outcome
- Alternatives they mention
- Trigger phrases they use when they are ready to switch or buy
If needed, read references/intent-signals.md.
5. Generate Search Queries
Create search queries for each platform using:
- - problem-first phrases
- comparison phrases
- recommendation phrases
- migration phrases
- complaint phrases
- workflow phrases
Prefer queries that look like what a founder would actually search manually. Include both broad and narrow variants.
6. Find Candidate Threads
Search for conversations where the author is:
- - actively asking for a tool
- comparing alternatives
- frustrated with an existing workflow
- requesting recommendations
- describing an expensive manual workaround
Do not treat every mention as an opportunity. Skip vanity mentions, generic news, and unrelated discussions.
7. Score Intent and Risk
Score each candidate thread across two dimensions:
- - Intent score: INLINECODE13
- Risk score: INLINECODE14
Use this rubric:
- - Intent
5: actively asking for a solution, recommendation, alternative, or migration path now - Intent
4: strong pain and explicit need, but not directly asking to buy - Intent
3: clear pain but vague urgency - Intent
2: adjacent discussion, weak buyer signal - Intent
1: mostly curiosity or news - Intent
0: not relevant
- - Risk
5: clear anti-promo context, likely rule violation, or obviously off-topic - Risk
4: probably poor fit, strong chance of backlash - Risk
3: mixed fit, needs caution - Risk
2: reasonable fit, still keep the reply light - Risk
1: low risk - Risk
0: very safe context
Do not recommend engagement on threads with intent lower than 3 or risk higher than 3 unless the user explicitly asks for aggressive exploration.
If the platform or community norms matter, read references/platform-guardrails.md.
8. Choose the Engagement Ladder
Choose one of three modes for each thread:
- -
Help only: answer the question without mentioning the product - INLINECODE31 : answer first, then lightly mention the product if it genuinely fits
- INLINECODE32 : recommend the product directly because the ask is explicit and the fit is strong
Default to the least promotional option that still serves the user.
9. Draft Replies in Native Style
Write like a participant in that platform, not like an ad.
- - Reddit: specific, plain, contextual, low-hype
- X: shorter, sharper, more conversational
Avoid:
- - generic praise
- vague hype
- repetitive language across threads
- fake neutrality when self-promoting
- pushing links before establishing relevance
If needed, read references/reply-patterns.md.
10. Produce a Decision-Oriented Output
Return the results in a format the user can act on immediately:
- - launch blockers if present
- market summary
- user summary
- competitor or substitute summary
- positioning recommendation
- primary positioning angle
- channel recommendation
- search queries
- top thread or account targets
- intent and risk scores
- recommended engagement mode
- draft reply
- rationale
- follow-up suggestion
Use the template in assets/outreach-output-template.md when the user wants a reusable report format.
Output Standards
When presenting opportunities, keep each item compact and decision-ready:
- - INLINECODE35
- INLINECODE36
- INLINECODE37
- INLINECODE38
When the user asks for many replies, vary the structure and opening sentence. Do not generate near-duplicates.
Guardrails
Treat these as hard constraints:
- - Do not optimize for volume over fit.
- Do not advise deceptive personas or fake customer identities.
- Do not recommend mass posting or mass replying.
- Do not ignore subreddit rules or platform rules.
- Do not present spammy behavior as “growth”.
- Do not force a product mention when a helpful answer alone is better.
When unsure, say the thread should be skipped.
Deliverables
This skill is strongest at producing:
- - repo-to-product-brief audits
- launch-readiness assessments
- lightweight market scans
- user pain summaries
- competitor and substitute maps
- positioning briefs
- community prospecting plans
- buyer-intent search queries
- scored lead/thread lists
- platform-native draft replies
- post idea banks based on repeated pain points
- “engage / wait / skip” decisions
- lightweight launch packs for developer-built products
Resources
Read only the resource that matches the task:
- -
references/intent-signals.md: use when building queries, judging buyer intent, or clustering pain-language - INLINECODE40 : use when researching users, competitors, substitutes, and category positioning
- INLINECODE41 : use when deciding whether engagement is safe on Reddit or X
- INLINECODE42 : use when deciding whether to lead with problem, audience, comparison, simplicity, trust, or workflow fit
- INLINECODE43 : use when drafting replies or post ideas
- INLINECODE44 : use when the user provides a repo, README, release page, or unfinished product positioning
- INLINECODE45 : use when the user wants a structured market or competitor scan
- INLINECODE46 : use when the user wants a concise positioning recommendation before outreach
- INLINECODE47 : use when the product brief is incomplete
- INLINECODE48 : use when turning a repo into a usable product brief and launch checklist
- INLINECODE49 : use when the user wants a reusable scoring worksheet
- INLINECODE50 : use when the user wants a structured report or operating cadence
Example Triggers
Use this skill for requests such as:
- - “Find Reddit threads where people need this app.”
- “Research X posts from people frustrated with this workflow.”
- “Score these threads and tell me which ones are worth replying to.”
- “Draft helpful Reddit comments that can mention my product if appropriate.”
- “Turn this product brief into community prospecting queries.”
- “I built this repo. Figure out how to market it.”
- “Audit this GitHub project before we try to promote it.”
- “Research the market for this product before suggesting marketing.”
- “Find similar tools and tell me how users talk about this problem.”
- “Figure out how this product should be positioned before we launch.”
- “Do a repo audit before launch.”
- “Run market research and competitor research for this app.”
- “Help me find users for my product.”
- “Create a launch strategy from this GitHub repo.”
- “Plan Reddit outreach for this product.”
- “Find customers for my app without spamming.”
社区需求挖掘
概述
将模糊的增长想法转化为可重复的工作流程:发现相关的社区讨论,评估意图,避开不合适的帖子,并起草以帮助为先、仅在适当时推广的回复。
默认采用人工审核机制。本技能用于需求捕获和社区挖掘,而非批量发帖或欺骗性自动化。
特别适用于:仓库审计、发布准备、市场研究、竞品研究、定位策略、寻找产品用户,以及规划Reddit或X平台的外联活动。
快速入门
在进行外联研究之前,先收集一份精简的产品简介。仅询问缺失的信息:
- - 产品名称和一句话描述
- 目标用户或买家
- 解决的主要痛点
- 最佳差异化优势
- 允许的主张和禁止的主张
- 期望的语气
- 目标链接、演示或行动号召
如果用户未提供简介,请使用assets/product-brief-template.md中的模板。
如果用户仅提供了仓库、落地页或README,请先进行发布准备审计和市场扫描,再进行社区挖掘。使用:
- - references/repo-launch-readiness.md
- references/market-research.md
- references/positioning-heuristics.md
- assets/repo-audit-template.md
- assets/market-scan-template.md
- assets/positioning-brief-template.md
操作顺序
除非用户明确要求更窄的输出,否则按此顺序执行技能:
- 1. 提取或完善产品简介。
- 如果输入是仓库、README或粗糙的落地页,则审计发布准备情况。
- 研究用户、竞品、替代品和品类格局。
- 选择一个主要定位角度。
- 将产品转化为用户的痛点语言。
- 生成搜索查询并定位候选帖子。
- 对每个帖子进行意图和风险评估。
- 选择参与阶梯。
- 以平台原生风格起草外联内容。
- 返回一份面向决策的报告。
具体示例
示例请求:
text
我构建了一个小型开发者工具。审计仓库,研究类似产品,找出最佳定位,然后在Reddit上找到人们正在寻找替代品的帖子,并起草我可以使用的回复。
示例输出格式:
markdown
发布障碍
市场摘要
- - 品类拥挤,有少数现有玩家
- 当设置过程过于缓慢时,用户会选择切换
定位
- - 主要角度:简洁性
- 一句话定位:为只需要X功能的开发者提供的轻量级替代方案
机会
- 1. 帖子:
- 意图:5
- 风险:2
- 模式:软提及
- 回复草稿:...
工作流程
除非用户只要求一个狭窄的输出,否则遵循此流程。
1. 构建产品简介
如果用户提供的是仓库而非完善的产品简介,则从以下内容推导简介:
- - README标题和副标题
- 安装步骤
- 发布资源
- 截图或演示资源
- 陈述的用例
- 明显的信任障碍,如未签名的二进制文件或缺失的文档
在推荐外联之前,指出发布障碍。常见障碍:
- - 无截图或演示
- 标题不够有力
- 受众不明确
- 缺少安装路径
- 信任摩擦
- 无明确的行动号召
- 定位过于狭窄
如有需要,请阅读references/repo-launch-readiness.md。
2. 研究市场和用户格局
在起草外联内容之前,了解品类:
- - 可能的用户是谁
- 他们雇佣该产品完成什么任务
- 他们已经提到的替代品或替代方案
- 市场是拥挤的、碎片化的还是服务不足的
- 最具差异化的角度是什么:速度、简洁性、价格、信任、控制、利基适配或工作流适配
研究类似产品时,捕获:
研究用户时,捕获:
- - 他们在哪里寻求帮助
- 他们如何描述痛点
- 触发他们切换的因素
- 反复出现的反对意见或信任问题
不要将市场研究简化为功能清单。优先考虑:
- - 定位差异
- 目标用户差异
- 信任和采用摩擦
- 发现渠道差异
如有需要,请阅读references/market-research.md并使用assets/market-scan-template.md。
3. 选择定位角度
在市场扫描之后,决定产品应以什么为先导。
基于证据而非偏好选择主要角度:
- - 问题:痛点明显且紧迫
- 受众:产品明确针对具有共同身份的狭窄群体
- 比较:用户已经在比较此品类中的产品
- 简洁性:现有产品感觉臃肿、复杂或昂贵
- 信任:用户担心失败、安全风险或设置摩擦
- 工作流适配:产品比通用工具更适合特定的重复性工作
选择一个主要角度,最多一个次要角度。
避免在同一信息中堆叠过多角度。一个狭窄但清晰的万能工具信息通常比宽泛但模糊的信息表现更差。
如有需要,请阅读references/positioning-heuristics.md并使用assets/positioning-brief-template.md。
4. 构建需求地图
将产品简介转化为真实用户描述问题的方式:
- - 挫败感
- 当前的变通方案
- 期望的结果
- 他们提到的替代品
- 当他们准备切换或购买时使用的触发短语
如有需要,请阅读references/intent-signals.md。
5. 生成搜索查询
为每个平台创建搜索查询,使用:
- - 问题优先的短语
- 比较短语
- 推荐短语
- 迁移短语
- 抱怨短语
- 工作流短语
优先选择看起来像创始人会手动搜索的查询。包括宽泛和狭窄的变体。
6. 寻找候选帖子
搜索作者正在进行的讨论:
- - 积极寻求工具
- 比较替代品
- 对现有工作流感到沮丧
- 请求推荐
- 描述昂贵的手动变通方案
不要将每个提及都视为机会。跳过虚荣提及、通用新闻和不相关的讨论。
7. 评估意图和风险
在两个维度上评估每个候选帖子:
使用此评分标准:
- - 意图 5:现在正在积极寻求解决方案、推荐、替代品或迁移路径
- 意图 4:强烈的痛点和明确的需求,但未直接询问购买
- 意图 3:明确的痛点但紧迫性模糊
- 意图 2:相关讨论,买家信号较弱
- 意图 1:主要是好奇心或新闻
- 意图 0:不相关
- - 风险 5:明显的反推广背景,可能违反规则,或明显离题
- 风险 4:可能不适合,有强烈反弹的可能性
- 风险 3:混合适配,需要谨慎
- 风险 2:合理适配,仍保持回复轻量
- 风险 1:低风险
- 风险 0:非常安全的背景
除非用户明确要求激进的探索,否则不建议在意图低于3或风险高于3的帖子上参与。
如果平台或社区规范很重要,请阅读references/platform-guardrails.md。
8. 选择参与阶梯
为每个帖子选择三种模式之一:
- - 仅帮助:回答问题而不提及产品
- 软提及:先回答问题,然后如果产品确实合适,轻描淡写地提及
- 直接提及:直接推荐产品,因为请求明确且适配性强
默认选择仍然服务于用户的最不具推广性的选项。
9. 以原生风格起草回复
像该平台的参与者一样写作,而不是像广告。
- - Reddit:具体、平实、有上下文、低炒作
- X:更短、更犀利、更对话化
避免:
- - 通用赞美
- 模糊的炒作
- 跨帖子的重复语言
- 自我推广时的虚假中立
- 在建立相关性之前推送链接
如有需要,请阅读references/reply-patterns.md。
10. 生成面向决策的输出
以用户可以立即行动的格式返回结果:
- - 发布障碍(如果存在)
- 市场摘要
- 用户摘要
- 竞品或替代品摘要
- 定位建议
- 主要定位角度
- 渠道建议
- 搜索查询
- 顶级帖子或账户目标
- 意图和风险分数
- 推荐的参与模式
- 回复草稿
- 理由
- 后续建议
当用户想要一个可重复使用的报告格式时,使用assets/outreach-output-template.md中的模板。
输出标准
在呈现机会时,保持每个项目紧凑且可立即决策:
- - 为什么这相关
- 为什么这有风险或安全
- 该说什么
- 为什么这种措辞合适
当用户要求多个回复时,改变结构和开头句子。不要生成近乎重复的内容。
护栏
将这些视为硬性约束:
-