AI Heartpower Coach

AI Heartpower Coach is a Codex skill for structured inner-clarity coaching, especially for founders and early-stage creators.

It helps users move from raw expression to clearer needs, decisions, energy signals, and small constructive next steps. It is not a therapy, medical, legal, or financial advice skill.

What It Does

The skill now starts with a lightweight current-state orientation when the user does not bring a clear concrete issue:

1. Ask 7 short option questions for a Seven-Dimension Heartpower Portrait.
2. Summarize the user’s current state.
3. Show the seven dimensions in text.
4. Generate a PNG structure chart when the environment supports file generation.
5. Recommend a user-facing coaching entry.
6. Enter the first coaching question.

If the user wants to skip the assessment, vent directly, or is emotionally flooded, the skill pauses the assessment and moves into structured listening instead.

Seven-Dimension Portrait

The startup portrait is not a psychological diagnosis. It is a routing layer that helps the user quickly see where to begin.

The seven dimensions are:

• 情绪混乱
• 需求模糊
• 心力不足
• 自我责备
• 决策压力
• 关系评价
• 角色拉扯

Each dimension is scored as an attention signal from 1 to 5. Higher means this part may need to be cared for first.

After the 7 questions, the skill generates a concise report:

当前现状：低电量状态 + 自我责备态

七维心力画像：
- 情绪混乱：低
- 需求模糊：低
- 心力不足：高
- 自我责备：中
- 决策压力：中
- 关系评价：中
- 角色拉扯：中

我先把这份结果整理成一张【当前心力结构图】，让你更直观看见七个维度现在的样子。

这会花几秒钟，生成好后我会先给你看图，再告诉你建议从哪里开始。

图生成好了。你可以先看一下这 7 个维度的分布。

我建议先从【整理心力消耗】开始。

User-Facing Entries

Internal mode names are useful for implementation, but the user sees plain-language coaching entries first:

PNG Portrait Generation

The skill includes a deterministic PNG renderer for the seven-dimension structure chart.

Example:

python3 scripts/render_portrait.py \
  --answers "1A2C3A4B5B6B7B" \
  --out "./generated-images/current-seven-heartpower-structure.png"

The generated image includes:

• Current-state title card
• Seven-dimension radar chart
• Seven score bars
• Short status descriptions
• Recommended user-facing entry

This renderer is intended for local Codex-like environments that can run Python and write files. In environments that cannot generate files directly, the skill should still provide the text portrait and can output the structured data for an external renderer.

The user-facing image path should be explicit:

1. Tell the user the chart is being generated and may take a few seconds.
2. Generate and display the PNG when supported.
3. Then explain the strongest signal and enter the recommended coaching path.

Core Modes

Basic inner-work modes:

• Release And Clear
• Need Discovery
• Experience Extraction
• Nonviolent Communication
• Need To Satisfying State

Founder-specific modes:

• Decision Fog
• Self-Doubt
• Opportunity Overload
• Energy Account
• User Demand Reality
• Founder Role Conflict

Mode Switching

Modes are not fixed labels. They are the current processing lens.

The skill checks each round whether the user’s center of gravity has changed. If a switch would help, it asks before switching:

我感觉我们现在可能已经从【当前模式】走到了【新模式】。
原因是你的重点从【旧问题】变成了【新问题】。
要不要切过去，专门处理这个问题？

Installation

Codex

Clone this repository into your Codex skills directory:

mkdir -p ~/.codex/skills
cd ~/.codex/skills
git clone https://github.com/zhaocathy1221-hash/ai-heartpower-coach.git

Then start a new Codex conversation. The skill should be available as:

$ai-heartpower-coach

If the skill does not appear immediately, restart Codex or open a fresh conversation.

Hermes Agent

Hermes Agent skills live in ~/.hermes/skills/. Clone this repository there:

mkdir -p ~/.hermes/skills
cd ~/.hermes/skills
git clone https://github.com/zhaocathy1221-hash/ai-heartpower-coach.git

Then start a new Hermes session and invoke it as a skill command:

/ai-heartpower-coach 帮我判断我现在处在哪个创业者心力模式。

You can also ask in natural language:

Use ai-heartpower-coach to help me identify my current founder mode.

OpenClaw

OpenClaw can install a Git-backed skill whose repository root contains SKILL.md:

openclaw skills install git:zhaocathy1221-hash/ai-heartpower-coach

To install from a specific release tag:

openclaw skills install git:zhaocathy1221-hash/[email protected]

If you prefer a manual install, clone this repository into a directory that OpenClaw scans for skills, then run:

openclaw skills check
openclaw skills list

If your OpenClaw setup uses custom skill roots, add the parent directory to skills.load.extraDirs in ~/.openclaw/openclaw.json.

Usage

Invoke the skill explicitly when you want structured listening instead of a direct task answer:

用 $ai-heartpower-coach 帮我判断我现在处在哪个创业者心力模式，然后只用这个模式陪我往下走。
我的充分表达是：……

To start with the seven-dimension assessment:

用 $ai-heartpower-coach 做一次当前心力画像测评。

To generate the image after answering:

生成当前心力结构图。

Use regular GPT/Codex directly when the task and desired output are already clear. Use this skill when the user is unclear, emotionally loaded, bodily resistant, or needs to understand themselves before producing work.

Example Prompts

用 $ai-heartpower-coach 帮我判断我现在处在哪个创业者心力模式，然后只用这个模式陪我往下走。
我的充分表达是：……

请进入机会过载模式。
我现在有几个方向，不知道该聚焦哪个。请帮我判断哪个最接近资源、哪个最容易验证、哪个最能积累长期资产、哪个可能只是焦虑驱动。

请先不要给建议。
先帮我把“我为什么这么抗拒这件事”可能包含的感受、身体信号、没被满足的需要说清楚。

Files

• SKILL.md: trigger description, routing rules, response principles, and mode switching behavior.
• references/startup-assessment.md: seven-question startup assessment, score mapping, report shape, user-facing entry names, and PNG generation guidance.
• references/modes.md: detailed mode definitions and mode-switch signals.
• references/expression-guide.md: sufficient expression prompts and body-signal listening guide.
• references/safety-boundaries.md: crisis, health, business, and dependency boundaries.
• scripts/render_portrait.py: local PNG renderer for the seven-dimension heartpower structure chart.
• agents/openai.yaml: Codex UI metadata.

Safety Boundaries

This skill is for reflective coaching and founder clarity. It should not diagnose mental or physical health conditions, replace professional care, or make major business decisions for the user.

For crisis, severe symptoms, or immediate danger, prioritize real-world support and local emergency resources.