PlugWiki

dbs-goal Skill for Claude: Turn a Vague Wish Into a Checkable Goal

dbs-goal, from dontbesilent's dbskill set, makes Claude turn a vague wish into a goal card with a deliverable and a finish line.

Installs
25.7K
on skills.sh
Stars
11K
on GitHub
Files
1
besides SKILL.md
Updated
Oct 2026
last repo commit

dbs-goal makes Claude read everything you have already said, pull out the deliverable, audience, scope, finish condition and limits, and ask a question only when the answer would change the result. If the goal is already clear it gives a short goal card and stops. If one thing blocks it, it states what is settled and asks that single question.

For a pure wish such as "I want a personal brand" or "I want to get better", it asks one narrowing question per turn, then hands back a checkable goal, finish conditions and one next step. It never picks numbers, deadlines or success criteria for you.

The SKILL.md is written in Chinese and tells Claude to answer in whatever language you use. It is one SKILL.md plus one agent config file, with no scripts. It belongs to the 34-skill dbskill set, but the SKILL.md does not require any other skill.

Good for

  • Founders and freelancers who start tasks with 'I want to build my brand' and need a concrete target
  • Anyone briefing Claude on a deliverable who keeps getting answers to the wrong question
  • Team leads who want a short, checkable brief before work starts
  • Chinese and English speakers: it replies in the language you write in

Install dbs-goal

Claude Code

  1. 1

    Copy this prompt.

    Install the dbs-goal skill in Claude Code. Run this command:

npx skills add https://github.com/dontbesilent2025/dbskill --skill dbs-goal

Then check that it loaded and tell me how to use it.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

提取已有约束，只追问影响执行或验收的信息，把模糊愿望整理成可行动、可检查的交付目标。用户要明确做成什么、边界和验收标准时使用；不替代概念拆解或研究问题改写。

Sections in SKILL.md

  1. 核心哲学
  2. 1．意义即使用
  3. 2．语义做工检测
  4. 3．家族相似性
  5. 4．保留用户主权
  6. 工作流程
  7. Step 0：读取全部上下文
  8. Step 1：选择处理路径
  9. 条件测试
  10. 可指物测试
  11. 完成边界测试
  12. 失败边界测试
  13. 语义做工测试
  14. 上下文兼容测试
  15. 提问纪律
  16. 输出

Key facts

Author
dontbesilent2025 (GitHub: dontbesilent2025)
Repo
dontbesilent2025/dbskill
License
CC BY-NC 4.0 (non-commercial; commercial use needs the author's permission)
Installs on skills.sh
25,732 as of 2026-11-02
GitHub stars (whole repo)
11,039 as of 2026-11-02
Last repo commit
2026-10-28
Files
2 files including SKILL.md (agents/openai.yaml), no scripts
skills.sh audits
Gen Agent Trust Hub pass, Socket pass, Snyk pass

What it includes

  • Extract what you already told it before asking anything

    It reads your message, earlier chat, attachments and confirmed facts, and fills a table: deliverable, audience, scope, specs, finish condition, limits, later use. It will not ask you to repeat what you gave.

    Step 0: read all context

  • Clear goal: output a short goal card and stop

    When the deliverable, next action and finish condition are known, it gives a goal card with deliverable, audience, scope, finish condition, assumptions and one next step.

    Try: “Write a 10-page PDF market report for our sales team, covering the 2025 EU data, due Friday.”

    Path A: already executable

  • States what is settled, asks one decision-changing question

    It lists what is clear, picks the single missing item that would change the deliverable, path or acceptance, and waits for your answer before re-judging.

    Path B: one blocking gap

  • One narrowing question per turn, then a checkable goal

    For wishes like 'build a personal brand', it runs the condition tests below, one question at a time, and ends with a checkable goal, finish conditions and next step.

    Try: “I want to build a personal brand.”

    Path C: the goal is only a wish

  • Asks what the step should change, only when it matters

    It asks what you want to change once this step is done only when the value depends on a later result or you revealed a higher aim. If that does not change the plan, it stops asking.

    Path D: goal may be only a means

  • Referent, finish line, failure line, word-work, fit with context

    Referent test: what could you point to when done. Finish-boundary test: what evidence ends the round. Failure-boundary test: what would make you say it failed. Word-work test: flags vague words like 'better' or 'systematic'. Context-fit test: which constraint is most likely to change the goal.

    Five condition tests

  • Four checks before it is allowed to ask

    It asks only if the answer is not already in the chat, cannot be safely inferred, would change the deliverable, path or acceptance, and brings new decision information. It may offer a low-risk assumption you can correct.

    Question discipline

  • Goal card, single question, or full goal audit

    Three outputs: the goal card, the 'what is settled plus one question' note, and for vague goals a goal audit with your quote, unclear terms, a checkable goal, finish conditions and next step. It skips the long audit unless you ask for a full report.

    Output formats

From the official docs

Questions

What does the dbs-goal skill do?

It turns a vague wish into a goal Claude can act on and check. It pulls constraints from what you already said, asks one question only when the answer would change the result, and returns a goal card with the deliverable, audience, scope, finish condition and next step.

How do I install dbs-goal in Claude Code?

Use the install box above, which installs this skill from the dbskill repo with the skills CLI. The README also documents a Claude Code plugin route for the whole set, with commands prefixed dbs:, but this page does not cover that route. After installing, ask for it by name, for example /dbs-goal, or describe a goal you want pinned down.

Does it work in English?

Yes. The SKILL.md is written in Chinese, but it tells Claude to reply in English when you write in English and in Chinese when you write in Chinese.

Does it need an API key or other tools?

No key is named in the SKILL.md. The folder holds the SKILL.md and an agents/openai.yaml file, and bundles no scripts. It works by changing how Claude reads your request and what it asks back.

Is dbs-goal safe to install?

It is a set of instructions: the skill folder has two files and no scripts. It tells Claude to extract constraints, ask questions and write a goal card. Read the SKILL.md before installing if you want to check it yourself. Note the licence is non-commercial (CC BY-NC 4.0 per the README).

Do I need the other dbskill skills?

No. The SKILL.md does not call other skills. The README sells the full 34-skill set with a /dbs router that picks skills for you, and a skill like dbs-goal works alone when installed on its own.

How is it different from just asking Claude to clarify my goal?

It sets rules for the questioning. It extracts what you already said, asks at most one question per turn, only when the answer changes the outcome, and refuses to invent numbers, deadlines or success criteria. It also keeps clear requests from being interrogated.

Information

updated Nov 2, 2026

Repository
dontbesilent2025/dbskill
Folder
skills/dbs-goal

dontbesilent2025/dbskillskills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 2
  1. dbs-goal SKILL.md · github.com
  2. dbskill README · github.com

Updated Nov 2, 2026.