跳到正文
cccskills
免费在 GitHub 公开

stss

Reduce defensive disclaimers, stacked hedging, and self-protective narration when drafting, rewriting, or auditing proposals, articles, case studies, captions, and other reader-facing prose. Use when defensive wording is requested for review or appears in a draft, including wording the assistant adds. Preserve meaningful limits; do not perform general style cleanup or ordinary code work.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(3)

  • SKILL.md4.6 KB
  • agents/openai.yaml311 B
  • references/examples.md7.4 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Stop That Shit Slop

别再废话。

Make the point without writing for an imaginary critic. Preserve limitations that change what the reader understands or decides.

Stay in scope

Reduce defensive wording in reader-facing prose, including captions and introductions. Do not classify text as AI-written, promise to remove an "AI voice," humanize unrelated prose, or perform a general style cleanup.

Judge sentences by their reader use, including sentences introduced during drafting. A request to describe a project's evolution calls for that history; it does not by itself call for an account of how the article's sources were checked.

Select the mode

  • rewrite: Return a tighter version. This is the default when the user provides text without a mode.
  • audit: Identify defensive writing and propose the smallest useful fix. Do not rewrite the full text.

Follow an explicit mode. Infer the mode from the request only when none is given.

For a drafting request, create the requested artifact from the supplied material and apply the rewrite checks before returning it.

Record the claims

Before editing, build a private Claim Ledger from the supplied material:

  • facts and numbers;
  • named actors and sources;
  • evidence strength;
  • stated uncertainty and scope;
  • what the reader needs to understand or do.

Do not print the ledger unless the user asks for it. Do not add facts, sources, numbers, certainty, or causality.

Keeping facts accurate does not require printing every editorial note. Keep verification logs and source-checking process in working notes unless the reader needs them. Preserve useful attribution, actual methods, and conditions that affect a claim.

Test each defensive sentence

Apply the Sentence Consumer Test:

  1. Who will use this sentence?
  2. What understanding or decision does it change for this reader?
  3. What becomes false or misleading if it is removed?

Choose one action:

  • DROP: It only narrates diligence, anticipates criticism, apologizes, or says what the document does not attempt.
  • CALIBRATE: It contains real uncertainty but uses stacked hedges. Keep one precise limitation.
  • RELOCATE: The reader needs it beside a particular claim, or only the editor needs it in working notes. Choose the destination by its consumer; moving an unnecessary disclaimer to a footnote does not make it useful.
  • KEEP: It changes the meaning, evidence strength, attribution, or a legal, safety, financial, methodological, contractual, or scope decision.

Do not decide from a keyword alone. The same phrase can be waste in one context and necessary in another.

Rewrite

  1. Lead with the requested subject, claim, decision, or proposal.
  2. State scope positively: say what the work covers.
  3. Replace stacked hedges with one evidence-matched qualifier.
  4. Remove internal process narration unless the process is requested or material.
  5. Put a necessary limitation beside the claim it limits.
  6. Check the complete artifact, including captions and assistant-added prefaces, against the Claim Ledger and Sentence Consumer Test. Do not reintroduce dropped defenses in the delivery note.

Return only the requested artifact unless the user asks for commentary. If the text is already direct and faithful, return it unchanged.

Audit

Report only actionable findings. For each finding, give:

  • the span or sentence;
  • DROP, CALIBRATE, RELOCATE, or KEEP;
  • the reader's interpretation or decision affected;
  • the smallest replacement when one is needed.

Use NO_DECISION_CONSUMER when a sentence serves neither reader understanding nor a decision. If there are no actionable findings, say so and stop.

Check the claim diff

Before returning the result, verify:

  • no supplied fact, number, source, or actor disappeared without reason;
  • no new fact, number, source, or actor appeared;
  • uncertainty did not become certainty;
  • correlation did not become causation;
  • a requested or meaning-changing limitation remains visible.

Run one claim-preserving pass. Do not add a score, a probability that text is AI-written, a change diary, or repeated self-review.

Read references/examples.md when the boundary is ambiguous or the user asks for examples.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Complete requested work while preventing speculative defenses and scope creep. Use when considering extra hardening, reviewing possible overengineering, resolving repeated verification loops, following explicit task boundaries, or when the user invokes Stop That Shit. Ordinary use of validation, retries, or dependencies alone is not a trigger.

中文简介正在准备中。当前显示原文说明。

lennney/stop-that-shit27802026年10月19日 更新

查看 lennney 的全部技能

报告此技能的问题