Workflow & Core Tools · dontbesilent2025/dbskill

dbs-jtbd

Name: dbs-jtbd
Author: dontbesilent2025/dbskill

dontbesilent JTBD Job Clarification. Uses the Jobs to Be Done framework to identify the progress users want to make in specific contexts, the forces driving solution switching, and observable choice criteria—optimizing products, content, services, decisions, and AI prompts accordingly. Invocation command: /dbs-jtbd. Use when users ask "What problem are we actually solving?", "Why choose this solution?", or "Rewrite this prompt using JTBD".

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit

Author dontbesilent2025GitHub dontbesilent2025/dbskill ↗Stars 10121License CC BY-NC 4.0（仓库根 LICENSE：Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)；署名 + 禁止商业使用；Copyright (c) 2026 dontbesilent）Commit 8b8e33f1ec

Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Pure prompt orchestration with zero scripts or external dependencies: All capabilities derive from the judgment framework, questioning discipline, and output templates in SKILL.md

skills/dbs-jtbd/SKILL.md

本 Skill 从 Jobs to Be Done 视角理解用户要完成的事，用于产品、内容、决策或 AI 协作中的任务澄清。

The directory contains only SKILL.md and agents/openai.yaml (interface metadata), with no scripts/references/data.

Core mechanism is "Progress first, solution second": Demotes user-requested deliverables (writing articles, creating courses, building agents) to candidate means, then reverse-engineers the progress they want to make

skills/dbs-jtbd/SKILL.md

用户说「帮我写一篇文章」「我需要一个课程」「给我做一个 Agent」时，先不要把这句话直接当成需求结论。它通常只说明了用户想到的方案。

Supported by a five-stage chain: Context → Blocked progress → Desired outcome → Current solution → Choice criteria.

Job statements are enforced into fillable syntactic templates requiring "progress written as change, outcomes grounded in perceptible states", filtering out vague adjectives from the output

skills/dbs-jtbd/SKILL.md

当我处于 `{情境}`，我想要 `{推进的进展}`，以便 `{得到的结果}`。

The same section establishes two negative examples: do not merely repeat the action (e.g., "organize interview notes"), and avoid vague phrases like "increase efficiency".

Uses JTBD's three-layer jobs and the "Four Forces of Progress" (Push, Pull, Anxiety, Habit) as a fixed checklist, requiring recommendations to address all four forces rather than merely magnifying selling points

skills/dbs-jtbd/SKILL.md

一个方案被采用，通常需要推力和拉力强过焦虑与习惯。输出建议时要处理这四种力量，不能只放大卖点。

The three-layer job table specifically highlights: functional jobs determine deliverables, while emotional and social jobs often dictate messaging, friction, and final choice.

"Minimal questioning" discipline: First drafts job hypotheses based on existing materials, asking at most 1 concrete factual question only when missing information would change recommendations, while forbidding broad, open-ended questions

skills/dbs-jtbd/SKILL.md

只有以下信息缺失且会改变建议时，才问 1 个最小问题：

The text provides a positive example: "Last time you tried to solve this, at which step did you get stuck?", contrasting with negative examples like "What is your pain point?" or "Who is your target audience?".

Choice criteria must be "observable, evaluable, and prioritized": Categorized into Must-haves / Value-adds / Acceptable trade-offs, requiring phrases like "simple and easy to use" to be unpacked into verifiable statements

skills/dbs-jtbd/SKILL.md

标准要可观察。把「简单好用」还原为「第一次使用 10 分钟内能否得到可修改的结果」这类表述。

This is the primary distinction between this skill and generic "user persona" rhetoric.

Outputs are routed by scenario: AI collaboration → rewritten prompt; Product or service → job definition and switching anxiety mitigation; Content sales → context and credible proof; Personal decisions → minimum viable verification action

skills/dbs-jtbd/SKILL.md

| 产品或服务 | 任务定义、雇用时刻、需求优先级、降低切换焦虑的设计 |

It also specifies: If the user only wants a single answer, copy snippet, or prompt, skip displaying the full framework and deliver directly after internal evaluation.

Built-in prompt skeleton for AI collaboration that front-loads the job statement and complements context, materials, boundaries, deliverables, and acceptance criteria—enabling models to derive solutions from constraints

skills/dbs-jtbd/SKILL.md

把任务陈述放在提示词开头，并补上情境、已有材料、边界、交付物和验收标准。

The prompt skeleton also includes a built-in self-correction clause at the end: "If the job misaligns with my proposed solution, point out the misalignment first".

2Core Capabilities

01Job clarification: Translates "I want an X" into a "When..., I want to make progress on..., so that..." job statement

02Surface request vs. job misalignment identification (retaining the user's original solution as a candidate without arbitrary dismissal)

03Three-layer job inspection (functional, emotional, social), outputting only the layers relevant to the current task

04Switching forces diagnosis (Push, Pull, Anxiety, Habit)

05Observable choice criteria extraction and priority tagging

06Scenario-based action recommendations (AI collaboration, products/services, content sales, personal decisions)

07JTBD prompt skeleton generation (front-loading the job statement at the start of prompts)

08Evidence discipline and anti-persona stance: Refuses demographic traits, industry tags, or product names as job evidence; prohibits fabricating interviews and behavioral data

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Can be auto-loaded

For Administrators

Evidence & Triggers (File Paths & Excerpts)

SKILL.md: Pure prompt instructions and rule text, without script execution or external side effects

Audit Notes: The evaluation target is this skill's own directory: It contains only SKILL.md and one interface metadata yaml, with no scripts, network endpoints, credential reading, or disk writes, qualifying as purely instruction-based. The main consideration is not code safety but "framework bias"—if the JTBD three-layer jobs and Four Forces matrix are applied mechanically, they may yield an overly grandiose job definition relative to the user's original problem; the skill mitigates this through rules such as "do not ask continuous questions just to fit the framework" and "do not force claims when evidence is missing".

5Second-Pass Verification

[ok] 外部资源/工具：是否存在 SKILL.md 未声明的依赖 — 目录内仅 SKILL.md 与 agents/openai.yaml；token 扫描无 URL/CLI/API/包名命中，external_deps 为空数组成立。agents/openai.yaml 只含 display_name/short_description/default_prompt 三个界面字段，非运行依赖。
[ok] 安全结论：是否真的零文件读写 — 全文无读取或写入指令——不要求读用户文件、不要求落盘产物（唯一「输出」是对话中的 markdown 模板）。相比同批 dbs-save/dbs-report/dbs-knowledge 有明确落盘动作，本 skill 属同批风险最低档。
[ok] 实现原理是否成立：宣称的「切换力量」与「可观察标准」是否真有可执行定义 — 四力表给出每力的具体提问句（如「旧做法造成了什么具体损失、压力或阻塞？」），选择标准要求三分类并标优先级、给出「简单好用→第一次使用 10 分钟内能否得到可修改的结果」的还原示例。非口号式条目。
[unlocatable] 证据定位：star/last_push 等仓库级元数据 — stars 与 last_push 来自站点池 data/pool.json 的 GitHub API 采集值，pin 源码内无此数据，无法在仓库文件内逐字命中；license 可在仓库根 LICENSE 逐字命中。

6Verdict & Scenarios

Transforms "clarifying the problem first" into a reusable evaluation chain rather than vague questioning: Five-stage chain (Context → Progress → Outcome → Solution → Criteria) + Four Forces matrix + Three-layer jobs, with concrete question prompts and output anchors for each

Highest practical utility for AI collaboration: Delivers ready-to-paste prompt skeletons with a built-in self-correction clause: "If the job misaligns with my proposed solution, point out the misalignment first"

Disciplined questioning: Does not ask follow-ups if materials are sufficient, posing at most 1 factual question only when it alters recommendations—avoiding the "questionnaire fatigue" common in consultative skills

Built-in evidence discipline: Distinguishes facts, inferences, and unconfirmed items; forbids fabricating interviews and behavioral data; writes hypotheses instead of making up stories when evidence is missing

Best for: Ideal for pre-execution requirement calibration: defining products or services, planning content topics, structuring personal decisions, or auditing self-written AI prompts—especially to resolve the common misalignment where "the user's proposed solution is not the real need." Also well-suited as an interview script for content creators and consultants to directly reuse the Four Forces matrix and Three-layer jobs as interview outlines.

Not for: Not suitable for scenarios where specifications are already clear and only execution remains (where follow-up questions only add unnecessary overhead); not suitable for business judgments requiring external data or validation (it does not perform search or verification); not suitable for workflows requiring persisted file deliverables—this skill does not generate any files, archives, reports, or planning documents (use dbs-save, dbs-report, or dbs-learning instead).

Install Agent-Ready

① Verified Mirror 2026-09-15

Option A · Direct DownloadDownload dbs-jtbd.tar.gz

sha256: 236334174ad98f29…

Option B · Copy Guide for Agent

install.json

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit 8b8e33f1ec.