needs-analysis
Mandatory high-priority need validation. Always use before general brainstorming or planning whenever Tony's message contains the Chinese word “需求” or the standalone English word “needs” (case-insensitive). Investigate product, customer, business, or software needs with supporting and disconfirming evidence; judge true, conditional, pseudo, or insufficient; score breadth, urgency, and frequency; then design a decisive validation before development.
- License 许可
- MIT
- 最近更新
- 2026-11-01
- Stars
- ★ 3
- Forks
- 1
01 前言 Front Matter
- name
- needs-analysis
- description
- Mandatory high-priority need validation. Always use before general brainstorming or planning whenever Tony's message contains the Chinese word “需求” or the standalone English word “needs” (case-insensitive). Investigate product, customer, business, or software needs with supporting and disconfirming evidence; judge true, conditional, pseudo, or insufficient; score breadth, urgency, and frequency; then design a decisive validation before development.
02 资源文件 2
- 参考文档agents/openai.yaml276 B
- 参考文档references/rubric.md3.9 KB
03 SKILL.md 正文
Needs Analysis
Turn a raw request into an evidence-backed product decision. Separate whether the problem is real from whether it is an attractive product opportunity.
Operating rules
- Start with the conclusion. Use
真需求,有条件成立,伪需求, or证据不足; include confidence (高/中/低). Do not force a binary verdict when evidence is missing. - Treat “广泛、刚需、高频” as opportunity qualities, not proof that a need exists. A need can be real but commercially weak.
- Consider a candidate attractive when the need is at least conditionally real and one or more of the three qualities scores at least 3/5. More passing qualities increase priority, but do not replace evidence.
- Distinguish facts, user claims, third-party estimates, inference, and unknowns. Cite or link sources when research tools are available.
- Prefer observed behavior over stated preference: payment, active workaround, repeated use, switching, budget, deadlines, support tickets, search or purchase behavior.
- Do not confuse a requested feature with the underlying need. Analyze the problem independently of the proposed solution.
- Continue on low-risk assumptions when possible. Ask only questions whose answers would materially change the verdict or validation test.
Read rubric.md before scoring or issuing the final verdict.
Workflow
1. Normalize the raw need
Rewrite it without solution bias:
[specific user]in[specific situation]struggles to[job/progress], causing[measurable cost or consequence]; today they use[current alternative/workaround].
Record separately:
- proposed solution;
- target user and buyer;
- triggering situation;
- desired outcome;
- current alternative;
- claimed frequency and consequence;
- uncertainties and assumptions.
If the user and situation are absent, infer a provisional version and label it as an assumption.
2. Build the evidence set
Search only as deeply as the decision warrants. Use this order:
- User-provided materials, current workspace, product analytics, customer notes, support records, sales or payment evidence.
- Direct customer evidence: interviews with concrete past behavior, paid pilots, preorders, retained usage, switching or active workarounds.
- Market behavior: competitor customers and pricing, reviews, complaints, procurement/RFPs, job posts, communities, search trends.
- Market reports, surveys, social engagement, and generic opinions as supporting proxies only.
For each important item capture: source, date, affected segment, what it supports or contradicts, and evidence strength. Recheck live sources for time-sensitive claims. If access is unavailable, state the limitation and do not fabricate findings.
Search for disconfirming evidence as deliberately as confirming evidence. Look for non-consumption, free substitutes, low retention, long replacement cycles, weak willingness to pay, or a segment too costly to reach.
3. Judge true versus false need
Evaluate these independent questions:
- Is there an identifiable user in a recurring or important situation?
- Does the problem exist without the proposed product?
- Is there observable cost, risk, delay, frustration, lost revenue, or missed progress?
- Do users already spend time, money, reputation, or effort on a workaround?
- Is the buyer able and willing to act now?
- Does contrary evidence weaken the claimed problem or segment?
Use the rubric's verdict rules. A verbal “I would use this” is weak evidence. A paid or repeated workaround is strong evidence.
4. Score the three opportunity qualities
Score each from 0–5 using the rubric and show one sentence of evidence:
- 广泛 Breadth: enough reachable users share the problem; define the denominator and reachable segment instead of saying “everyone”.
- 刚需 Urgency: inaction creates a material, time-bound consequence; distinguish must-have from useful.
- 高频 Frequency: the problem or workflow recurs often enough to support habit, retention, or repeat purchase.
Mark each quality 通过 at 3–5 and 未通过 at 0–2. Do not hide the three scores inside a single average.
Map passed qualities to product logic:
- 广泛: emphasize simple onboarding, distribution, standardization, or lower unit price.
- 刚需: emphasize outcome certainty, speed, trust, service, and premium pricing.
- 高频: emphasize workflow integration, retention, automation, and subscription or repeat use.
5. Design the smallest decisive validation
Identify the riskiest assumption and propose a test that can fail. Include:
- target participant and recruitment channel;
- artifact or offer;
- behavior to observe;
- sample size or exposure;
- success threshold;
- failure/stop threshold;
- timebox and next decision.
Prefer commitment tests over opinion tests: paid diagnostic, deposit, signed pilot, data access, calendar commitment, or repeated usage. For Tony's AI 本机工作台服务, when relevant, map the test to the current 50 触达 / 10 访谈 / 3 demo / 1 付费意向 gate and recommend the next missing external action.
6. Deliver the analysis
Use this structure, scaling detail to the decision:
- 结论 — verdict, confidence, and one-sentence reason.
- 需求重述 — user, situation, job, consequence, current alternative.
- 证据与反证 — compact table with source/date/type/strength/implication.
- 真伪需求判断 — behavioral evidence, gaps, and why the verdict follows.
- 三要素评分 — 广泛 X/5, 刚需 X/5, 高频 X/5; pass count and product implications.
- 产品决策 — build now, validate first, narrow segment, reposition, or stop.
- 最小验证实验 — exact test, thresholds, timebox, and next action.
- 仍未知 — only uncertainties capable of changing the decision.
When the user also requests implementation, present the analysis first, convert validated findings into acceptance criteria, then continue the implementation unless a failed gate makes building wasteful or unsafe.
Failure modes
- Do not call a need real solely because many people liked, searched, or discussed it.
- Do not call a need false solely because it is niche or infrequent; urgent niche problems can support premium products.
- Do not count the same evidence twice across breadth, urgency, and frequency.
- Do not use market size as a substitute for reachable users or willingness to pay.
- Do not make interview count the goal; use interviews to uncover past behavior and then seek commitment.
- Do not recommend a full build when a landing page, manual concierge test, paid diagnostic, or prototype can test the riskiest assumption faster.
04 同仓库更多 Skills
seedance-prompt-en
Write effective prompts for Jimeng Seedance 2.0 multimodal AI video generation. Use when users want to create video prompts using text, images, videos, and audio inputs with the @ reference system. Covers camera movements, effects replication, video extension, editing, music beat-matching, e-commerce ads, short dramas, and educational content.
onepage-pdf
Convert an HTML page into a single continuous-page PDF (one tall page, no pagination breaks), preserving desktop layout, backgrounds and selectable text. Use when the user asks to turn an HTML file/report/proposal into a "single-page PDF", "long PDF", "不分页 PDF", "单页 PDF", "长图式 PDF", or complains that an HTML-to-PDF export breaks into pages or cuts content. Supports optional string redaction with leak verification.
native-feel-cross-platform-desktop
Use when the user is designing, prototyping, or rewriting a desktop app that must run on multiple OSes (macOS + Windows, optionally Linux) AND feel indistinguishable from a native app to its users — fast launch, native windowing, native input handling, native materials. Trigger words include "cross-platform desktop", "Electron alternative", "Tauri vs native", "WebView wrapper", "near-native performance", "Raycast architecture", "WebKit/WebView2 quirks", "WKWebView", "system tray app", "global hotkey app", "launcher app". Do NOT trigger this skill for pure web apps, pure mobile apps, or for greenfield projects that have no native-feel requirement.
gstack
Router for the gstack skill suite. (gstack)
fireworks-tech-graph
Use when the user wants to create any technical diagram - architecture, data flow, flowchart, sequence, agent/memory, or concept map - and export as SVG+PNG. Trigger on: "画图" "帮我画" "生成图" "做个图" "架构图" "流程图" "可视化一下" "出图" "generate diagram" "draw diagram" "visualize" or any system/flow description the user wants illustrated.
dbs-video-extract
接收抖音、小红书、微信视频号等短视频链接或分享文案,查询作品/账号数据并提取语音文字稿,按作者和标题归档为 Markdown。用户希望完整解析短视频、取得数据或文字稿时使用;运行前检查 TikHub 与轻抖使用凭证,缺少时用新手能理解的方式引导购买和安全配置。