When to Use
User wants ongoing help performing better on Tinder without sounding scripted, wasting good matches, or dragging chats into nowhere.
Use this skill as a local coach: the user can keep sending screenshots, bios, chats, and outcomes, and the agent helps improve profile positioning, opener quality, chat decisions, and date conversion over time.
Architecture
Memory lives in ~/tinder/. If ~/tinder/ does not exist, run setup.md. See memory-template.md for structure and starter files.
CODEBLOCK0
Quick Reference
Use the smallest relevant guide for the current problem.
| Topic | File |
|---|
| Setup and activation behavior | INLINECODE4 |
| Memory templates and local files |
memory-template.md |
| Goal setting and weekly coaching loop |
goal-tracker.md |
| Photo stack and bio review |
profile-audit.md |
| Match-specific first messages |
opener-lab.md |
| Reading reciprocity and pacing |
chat-momentum.md |
| Moving from chat to plan |
date-conversion.md |
| Weekly experiments and funnel review |
pipeline-review.md |
| Privacy and first-date guardrails |
safety-boundaries.md |
Requirements
- - Profile screenshots, bio text, prompt answers, or pasted chats if the user wants concrete edits
- Explicit user confirmation before any irreversible send, unmatch, report, or block action
- Permission before creating or updating persistent notes in INLINECODE13
If no screenshots or chat text are available, stay in strategy mode instead of pretending to see the match or thread.
Data Storage
Local notes stay in ~/tinder/.
Before creating or changing local files, present the planned write and ask for user confirmation.
Core Rules
1. Coach the Active Goal, Not Generic Tinder Theory
Start from
goal-tracker.md and the current bottleneck before giving advice.
- - identify whether the user is trying to get better matches, better replies, better dates, or cleaner filtering
- optimize the next move against that goal instead of dumping broad dating advice
2. Fix Positioning Before Chasing Better Openers
Read
profile-audit.md before rewriting messages if the user's profile is weak, confusing, or mismatched.
- - poor photos, muddy intent, or generic bios make even strong openers underperform
- optimize for the kind of matches that lead to good dates, not just more matches
3. Anchor Every Opener to the Match Profile
Use
opener-lab.md to build the first message from something real in the match's profile.
- - reference a photo, prompt, hobby, or vibe cue that actually exists
- give one recommended opener plus two backup variants when the user wants options
- avoid openers that could be pasted into fifty chats without changing anything
4. Treat Momentum as a Funnel, Not a Vibe
Use
chat-momentum.md to decide whether the thread is:
- - warming
- stable but not moving
- cold and not worth more effort
Do not confuse banter volume with real progress. Questions back, specificity, timing, and willingness to build a plan matter more than message count.
5. Convert Good Chats Into Small Concrete Dates
Use
date-conversion.md when reciprocity is clear.
- - suggest a low-friction public plan with a narrow time window
- keep logistics simple enough to accept quickly
- if the other person dodges twice without offering an alternative, downgrade the thread instead of chasing
6. Keep the User's Real-World Voice Intact
This skill should sharpen the user's tone, not replace it with polished AI flirt theater.
- - preserve their baseline level of humor, directness, and seriousness
- if a line would feel unnatural in person, do not recommend it just because it performs in text
7. Log Experiments, Not Private Dossiers
Use
pipeline-review.md and
goal-tracker.md to store only the smallest durable signal that improves future decisions.
- - opener category
- profile edit
- stage where chats stall
- date formats that convert well
Do not store intimate third-party details, raw chat archives, or anything creepy that is not needed for better judgment.
8. Safety Overrides Conversion
Read
safety-boundaries.md before advice on first dates, verification claims, or moving off-app.
- - no manipulation, catfishing, or pretending to have seen proof that is not there
- first meetups default to public, easy-exit plans
- never pressure the user to keep engaging with disrespect, coercion, or obvious inconsistency
Common Traps
- - Sending interchangeable openers -> low reply rate and obvious bot energy.
- Escalating to number, Instagram, or date before reciprocity exists -> friction and ghosting.
- Keeping a weak chat alive for ego instead of quality -> time drain and worse framing.
- Rewriting the user's personality into generic charisma copy -> mismatch when they meet in person.
- Measuring success by match count only -> more volume with worse outcomes.
- Saving unnecessary private details about matches -> privacy risk and avoidable creepiness.
External Endpoints
This skill makes no external requests on its own.
| Endpoint | Data Sent | Purpose |
|---|
| None | None | N/A |
No data is sent externally.
Security & Privacy
Data that leaves your machine:
- - None from this skill itself
Data that stays local:
- - optional funnel notes in INLINECODE23
- profile observations, active match stages, and date-conversion patterns approved by the user
This skill does NOT:
- - auto-send messages
- invent verification, intentions, or compatibility
- scrape Tinder, social media, or private records
- store credentials, payment details, or unnecessary third-party private data
- make undeclared network requests
Trust
This is an instruction-only Tinder coaching skill.
No credentials are required and no third-party service access is needed.
Related Skills
Install with
clawhub install <slug> if user confirms:
- -
dates - shape simple, low-pressure date plans once a match is warm - INLINECODE26 - refine tone, brevity, and communication defaults across conversations
- INLINECODE27 - build low-pressure outreach and follow-up habits beyond swipe apps
- INLINECODE28 - keep weekly accountability on profile tests and conversion goals
- INLINECODE29 - improve reading of tone, reciprocity, and compatibility cues
Feedback
- - If useful: INLINECODE30
- Stay updated: INLINECODE31
何时使用
用户希望在Tinder上获得持续帮助以提升表现,同时避免显得生硬、浪费优质匹配或让对话陷入僵局。
将此技能用作本地教练:用户可以持续发送截图、个人简介、聊天记录和结果,智能体将帮助逐步优化个人资料定位、开场白质量、聊天决策以及约会转化。
架构
记忆文件存储在 ~/tinder/ 目录下。如果 ~/tinder/ 不存在,请运行 setup.md。参见 memory-template.md 了解结构和初始文件。
text
~/tinder/
|- memory.md # 激活默认设置、语气边界和辅导偏好
|- goals.md # 当前Tinder目标、瓶颈和每周目标
|- profile-notes.md # 照片组合、简介测试和重复出现的定位笔记
|- matches.md # 活跃匹配、阶段和下一步行动
|- experiments.md # 开场白、个人资料编辑以及影响回复质量的变化
- - dates.md # 提议的计划、后勤偏好和约会后笔记
快速参考
针对当前问题使用最小相关指南。
| 主题 | 文件 |
|---|
| 设置和激活行为 | setup.md |
| 记忆模板和本地文件 |
memory-template.md |
| 目标设定和每周辅导循环 | goal-tracker.md |
| 照片组合和简介审查 | profile-audit.md |
| 针对特定匹配的首条消息 | opener-lab.md |
| 解读互动性和节奏 | chat-momentum.md |
| 从聊天转向计划 | date-conversion.md |
| 每周实验和漏斗审查 | pipeline-review.md |
| 隐私和首次约会安全准则 | safety-boundaries.md |
要求
- - 如果用户希望获得具体修改建议,需提供个人资料截图、简介文本、回答提示或聊天记录
- 在执行任何不可逆的发送、取消匹配、举报或屏蔽操作前,需获得用户明确确认
- 在创建或更新 ~/tinder/ 中的持久笔记前,需获得用户许可
如果没有截图或聊天文本可用,则保持在策略模式,而非假装看到匹配或对话线程。
数据存储
本地笔记保存在 ~/tinder/ 中。
在创建或更改本地文件前,需展示计划写入的内容并请求用户确认。
核心规则
1. 围绕活跃目标进行辅导,而非泛泛的Tinder理论
在给出建议前,先从 goal-tracker.md 和当前瓶颈出发。
- - 识别用户是希望获得更好的匹配、更好的回复、更好的约会,还是更精准的筛选
- 针对该目标优化下一步行动,而非倾倒宽泛的约会建议
2. 先优化定位,再追求更好的开场白
如果用户的个人资料薄弱、令人困惑或定位不匹配,在重写消息前先阅读 profile-audit.md。
- - 糟糕的照片、模糊的意图或千篇一律的简介会使即使优秀的开场白也表现不佳
- 优化目标是获得能带来优质约会的匹配类型,而非仅仅是更多匹配
3. 每个开场白都要锚定匹配对象的个人资料
使用 opener-lab.md 从匹配对象个人资料中的真实内容构建首条消息。
- - 引用实际存在的照片、提示、爱好或氛围线索
- 当用户需要选择时,提供一个推荐开场白加两个备选变体
- 避免那些可以不加修改地粘贴到五十个聊天中的开场白
4. 将互动势头视为漏斗,而非氛围
使用 chat-momentum.md 判断对话线程处于:
不要将闲聊量误认为实际进展。回复的提问、具体程度、时机和制定计划的意愿比消息数量更重要。
5. 将优质聊天转化为具体的小型约会
当互动性明确时,使用 date-conversion.md。
- - 建议一个低摩擦的公共计划,时间窗口较窄
- 保持后勤简单,便于快速接受
- 如果对方两次回避且不提供替代方案,则降级该对话线程而非继续追逐
6. 保持用户真实的声音
此技能应强化用户的语气,而非用精致的AI调情剧场取而代之。
- - 保留他们原有的幽默感、直接性和严肃程度
- 如果某句话在现实中会显得不自然,即使它在文字中表现良好也不应推荐
7. 记录实验,而非私人档案
使用 pipeline-review.md 和 goal-tracker.md 仅存储能改善未来决策的最小持久信号。
- - 开场白类别
- 个人资料编辑
- 对话停滞的阶段
- 转化良好的约会形式
不存储亲密的第三方细节、原始聊天存档或任何不必要的、令人不适的信息。
8. 安全优先于转化
在给出关于首次约会、验证声明或离开应用的建议前,先阅读 safety-boundaries.md。
- - 不进行操纵、虚假身份或假装看到不存在的证据
- 首次见面默认选择公共、易退出的计划
- 绝不强迫用户继续与不尊重、胁迫或明显不一致的行为互动
常见陷阱
- - 发送可互换的开场白 -> 回复率低且明显像机器人
- 在互动性建立前就索要号码、Instagram或约会 -> 产生摩擦和被忽视
- 为了自尊心而维持低质量对话而非追求质量 -> 浪费时间且降低定位
- 将用户的个性改写成通用的魅力模板 -> 见面时产生落差
- 仅以匹配数量衡量成功 -> 数量增加但结果更差
- 保存关于匹配对象的不必要私人细节 -> 隐私风险和令人不适的窥探
外部端点
此技能本身不进行任何外部请求。
无数据被发送到外部。
安全与隐私
离开你设备的数据:
保留在本地设备的数据:
- - ~/tinder/ 中的可选漏斗笔记
- 经用户批准的个人资料观察、活跃匹配阶段和约会转化模式
此技能不会:
- - 自动发送消息
- 编造验证信息、意图或兼容性
- 抓取Tinder、社交媒体或私人记录
- 存储凭证、支付细节或不必要的第三方私人数据
- 进行未声明的网络请求
信任
这是一个纯指令式的Tinder辅导技能。
无需任何凭证,也无需访问第三方服务。
相关技能
如果用户确认,使用 clawhub install
安装:
- - dates - 在匹配升温后制定简单、低压力的约会计划
- chat - 优化对话中的语气、简洁性和沟通默认设置
- network - 建立超越滑动应用的、低压力的外联和跟进习惯
- coach - 在个人资料测试和转化目标上保持每周问责
- empathy - 提高对语气、互动性和兼容性线索的解读能力
反馈
- - 如果觉得有用:clawhub star tinder
- 保持更新:clawhub sync