360 Secure SkillHub
360 安全技能中心

SimpliXio Loop

该 Skill 是纯提示词驱动的决策整理框架:先收集上下文,区分核心信号、邻近信号与噪声,再输出目标、最多 3 个优先级(含原因、行动、预期结果、置信度)、应忽略项及风险、反馈循环和一个小而可逆的下一步。另含周回顾模式与基于真实产出的产品营销草稿模式。

ph-7
ph-7
热度 41.32
最近更新 2026-05-11 17:44:36
分类:效率依赖:内部调用大模型 / 读写文件应用场景:任务管理 / 工作流 / 报告撰写 / 摘要总结 ...

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add simplixio-decision-loop
Metadata
name
simplixio-decision-loop
description
Turn messy work, research, product ideas, and codebase context into 3 priorities, clear actions, ignored noise, and a weekly review loop.
version
0.1.0
homepage
https://github.com/pH-7
user-invocable
true
metadata
{"openclaw":{"emoji":"🎯","homepage":"https://github.com/pH-7","always":true}}

SimpliXio Decision Loop

Use this skill when the user needs to reduce messy context into clear execution.

This skill is inspired by SimpliXio: a decision system that turns noise into 3 priorities.

Core Promise

Convert noise into:

  1. what matters
  2. why it matters
  3. what to do next

When To Use

Use this skill for:

  • project planning
  • startup/product decisions
  • AI agent workflows
  • codebase next steps
  • weekly reviews
  • launch preparation
  • marketing automation planning
  • reducing overwhelming notes, research, or tasks

Do not use this skill for:

  • pretending certainty where evidence is weak
  • generating fake traction, revenue, users, or metrics
  • broad motivational writing
  • public posting without explicit approval
  • replacing legal, financial, medical, or immigration advice

Operating Principle

Do not serve more. Serve better.

The goal is not to produce a long plan. The goal is to produce a decision-ready brief.

Workflow

1. Gather Context First

Before recommending anything:

  • inspect available files when relevant
  • read existing project docs
  • check current code paths
  • identify the active objective
  • identify constraints and blockers
  • avoid asking questions if the answer can be discovered

If context is missing but not blocking, make a clear assumption and continue.

2. Separate Signal From Noise

Classify inputs into:

  • Core signal: directly affects the goal
  • Adjacent signal: useful later, not now
  • Noise: distracts, duplicates, or creates low-value work

Never treat all information as equal.

3. Produce Exactly 3 Priorities

Return at most 3 priorities.

Each priority must include:

  • title
  • why it matters
  • next action
  • expected result
  • confidence level

If there are fewer than 3 real priorities, return fewer. Do not pad weak priorities.

4. Make Ignored Noise Visible

Always include:

  • what to ignore
  • why to ignore it
  • what risk exists if it is ignored

Ignored work is part of the value.

5. Add a Feedback Loop

For every priority, define how the user should judge whether it worked.

Use simple feedback:

  • useful / not useful
  • acted / not acted
  • keep / change / drop

6. Recommend The Next Small Move

End with one action that can be done now.

The final action should be:

  • concrete
  • small
  • reversible
  • useful within 30 to 120 minutes

Output Format

Use this structure:

Decision Brief

Goal

State the goal in one sentence.

3 Priorities

1. Priority title

Why: Action: Expected result: Confidence:

2. Priority title

Why: Action: Expected result: Confidence:

3. Priority title

Why: Action: Expected result: Confidence:

Ignore For Now

  • Item: Reason: Risk:

Feedback Loop

  • Useful / not useful:
  • Acted / not acted:
  • Keep / change / drop:

Do Next

One concrete next action.

Weekly Review Mode

Use Weekly Review Mode when the user asks for a weekly summary, weekly review, review loop, or what to do next week.

Weekly Review output:

Weekly Review

What Repeated

List recurring priorities, blockers, or signals.

What Mattered

List the strongest signals from the week.

What To Ignore Next Week

List recurring noise.

What To Build Next

Give 1 to 3 next moves.

One Decision For Next Week

State the single highest-leverage decision.

Marketing Automation Mode

Use Marketing Automation Mode when the user asks to turn product output into marketing.

Rules:

  • anchor every post in real product output
  • never invent traction
  • never invent users or revenue
  • avoid hype
  • avoid generic AI wording
  • prefer proof, lessons, and shipped progress

Marketing output:

Product Proof

What actually happened.

Angle

The strongest story angle.

Drafts

X

Short post.

LinkedIn

Short professional post.

Blog / Dev.to

Short draft outline.

Quality Gate

Pass / fail with reasons.

Tone

Be:

  • direct
  • calm
  • practical
  • specific
  • product-minded

Avoid:

  • hype
  • filler
  • generic advice
  • long unfocused plans
  • fake certainty

Attribution

Created by Pierre-Henry Soria, builder of SimpliXio.