Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Gingiris-1031/gingiris-skills --skill gr-user-interviewgit clone --depth 1 https://github.com/Gingiris-1031/gingiris-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/gingiris-1031/gingiris-skills/gr-user-interview)<a href="https://agentmods.dev/skills/gingiris-1031/gingiris-skills/gr-user-interview"><img src="https://agentmods.dev/badge/skills/gingiris-1031/gingiris-skills/gr-user-interview/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/gingiris-1031/gingiris-skills/gr-user-interview"><img src="https://agentmods.dev/badge/skills/gingiris-1031/gingiris-skills/gr-user-interview.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00079 | $0.02221 |
| Opus 5.5 | $0.00032 | $0.00888 |
| Sonnet 5.5 | $0.00016 | $0.00444 |
| Haiku 4.5 | $0.00008 | $0.00222 |
Grade A, and why
gr-user-interview scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 172 lines — stays where its author put it; the contents beside it link to each section on GitHub.
⚠️ 2C 产品的渠道调整
本 skill 默认 dev / B2B 渠道。2C 消费品 / 教育 / 应用：获客主战场是 垂类社区 + 短视频 + 垂直 KOL，按地区公开数据选第一平台（如印尼/泰国短视频已反超 Facebook）。KOL 优先 nano / micro 垂类——粉丝越多互动率越低，micro > mega 性价比更高。完整 2C 渠道数据库 + 公开来源见 → gingiris-seo-geo/references/2c-adaptation.md。
gr-user-interview — 用户访谈
什么时候用
- "我要找 PMF，怎么做访谈"
- "用户反馈很杂，不知道抓哪个"
- "我采访了 5 个人，都没 insight"
- "怎么避免用户说假话"
- "不知道该访谈谁 / 约不到人"
- "激活率、付费率该怎么定义和对标"
从访谈到上线验收
访谈之后需要与产研协作时，读取 需求交接与回访模板：分开原始证据和拟议方案，按基础能力、onboarding、商业化归类，给出优先级理由、负责人、可观察的验收结果和回访对象。bug 与 feature request 使用不同视图，不能把收集到的意见直接当成交付承诺。
核心原则（5 条）
-
问 behavior，不问 opinion ❌ "你会用这个功能吗？" → 会得到礼貌性"yes" ✅ "上次遇到 X 问题是什么时候？你当时做了什么？" → 得到真实行为
-
20 分钟访谈 >> 5 个 feature 问题 深度一个场景，比问 5 个浅问题更有价值
-
录音 / 整稿 > 当场记笔记 当场记会错过微表情和语气
-
5 人后暂停，复盘 不要连访 20 人才总结，5 人后大概率已看到模式
-
用词镜像 用户用什么词，你也用什么词 —— 不要替换成产品语言
访谈结构模板（20-30 分钟）
| 阶段 | 时长 | 问什么 |
|---|---|---|
| Warm-up | 2 min | "你今天做了什么？"（开放式，松弛下来） |
| Discover | 5 min | "上次做 X 是什么时候？能讲讲吗？" |
| Deep-dive | 10 min | "当时你用了什么工具？怎么用的？哪里卡住？" |
| Alternative | 5 min | "有没有试过其他方法？为什么没继续？" |
| Wrap-up | 3 min | "如果有一个魔法能解决这个问题，它会怎么工作？" |
Token Top10 访谈法（招牌打法）
后台用量 Top10 拉 list → 共享屏幕访谈（必录屏）→ transcript 喂 AI 提 feature/bug 进 Linear → 次日回访。每天 ≥2 场。
- 按 token 消耗 / 用量排 Top10，每周刷新——重度用户的反馈密度最高
- 必须共享屏幕 + 录屏：看真实操作 > 听口述
- Transcript 交给 AI 自动提取 feature request / bug，直接进 issue 工具
- 次日回访"你提的 X 已修/已排期"——反馈闭环就是留存杠杆
- 访谈后截 5-10 分钟关键片段给产研
邀约节奏：
- 1/3/7/14 天四次触达（Day 1 → 3 → 7 → 14 各跟进一次）
- 双通道：邮件模板 + 邮箱反查 LinkedIn 私信，邮件没回转 LinkedIn 补一轮
- 话术用宽泛理由（"早期团队想做访谈"），不暴露真实筛选原因——避免用户表演
组合技：
- KOL 访谈二合一：合作 YouTuber 顺便做用户访谈 + 可用性测试，一次合作两份价值（20-30 分钟视频访谈 ROI 最高）
- Wisprflow 打法：launch 前人工 onboard 前 500 用户逐个访谈，用势能带 launch
激活基准速查
| 项 | 基准 |
|---|---|
| 激活定义 | 必须是行为阈值：完成 ≥2 任务 / 连续 3 天 ≥20 轮对话 / 使用超 5 分钟——不是"登录过" |
| 注册 → 付费合格线 | >5%（Pro-C 5-8%、SMB 10-15%） |
| 流失定义 | 两次对话后 24h 不回访 = 流失 → 自动触发挽回邮件 |
| 付费归因 | 四时间戳交叉：注册 × 激活 × 支付 × 最近登录 |
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago Changed · +14 lines 082c99277f1c
- 14d ago Changed · +4 lines · +21 tokens per session 7f48005796f2
- 1mo ago First seen · 154 lines · 58 tokens per session scan A 3aa65a595916
gr-user-interview is a skill published in the GitHub repository Gingiris-1031/gingiris-skills (87 stars, last pushed 8d ago), licensed MIT. It adds 79 tokens to every session and 2,221 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
pmf-advisor
You are a product-market fit advisor with a mandate to challenge, not validate. Your default posture is skepticism. You have seen founders mistake good marketing for PMF, mistake a vocal minority for the market, and mistake retention of the wrong customers for product success. You apply rigorous frameworks and refuse…
user-research
User personas, customer journey maps, interview guides, usability testing, and card sorting. Use when building user understanding, mapping customer experiences, planning user research sessions, or defining Jobs-to-Be-Done.
first-customer-finder
Find evidence-backed potential first customers, beta users, or design partners from public demand signals. Use for product-specific prospect research, qualified shortlists, source-grounded outreach drafts, and follow-up searches refined by the user's feedback and local prospect history. Does not send outreach…
jtbd-extractor
Turn raw research into Jobs-to-be-Done statements showing what users are really trying to accomplish. Use when: extract jobs, jtbd analysis, jobs to be done, what job is the user hiring, underlying user needs.
apply-product-market-fit-test
Use when deciding whether a startup has found product-market fit before investing in growth, sales, or scaling — survey active users with Sean Ellis's "how would you feel if you could no longer use this product" question and treat 40%+ answering "very disappointed" as the threshold, rather than relying on founder…
assumption-mapper
Surfaces the assumptions a product or strategy depends on, classifies them by criticality and evidence quality, and produces a prioritized test plan. Use when: assumption mapping, identify assumptions, what could go wrong, leap-of-faith assumptions, riskiest assumption test, RAT, assumption inventory, validate before…