DeepSeek Harness Remote
DeepSeek Harness Remote 可将手机或另一台电脑与你运行 DeepSeek Harness 的电脑安全连接,随时随地继续你的工作,方便随时查看进度并继续工作。
翻譯準備中
個人 GitHub 帳號下公開儲存庫的擁有者可以登入並驗證所有權,自行管理介紹,無需管理員核准。
登入並產生驗證檔案,在 30 分鐘內將其提交至此儲存庫的預設分支。其他分支或分叉儲存庫中的檔案無效。然後在下方檢查檔案,即可無需管理員核准直接管理專案。請保留檔案以維持管理權限。
如需登入,將前往登入頁面。登入後請回到此處,再次點擊所選的驗證按鈕。
在專案資料夾中執行 /nomorevibe verify,即可驗證所有權並自行管理介紹。
一分鐘完成驗證Free Claude skill that turns rough requests into clear, complete prompts using a six-part framework (task, context, audience, format, example, avoid).
翻譯準備中
取自儲存庫 README
正在載入完整 README。
檢視完整 README ↗A free Agent Skill that teaches Claude to turn a rough request, or a weak prompt, into a clear and complete one, using a simple six-part framework.
Ask Claude things like:
Claude finds the missing pieces, asks at most a few short questions (or uses [placeholders] if you want speed), and hands back a copy-ready prompt with a short note on what changed.
| Part | What it answers |
|---|---|
| 1. Task | What exactly should the AI produce? |
| 2. Context | What background can't the AI know? |
| 3. Audience and purpose | Who reads it, and what should happen next? |
| 4. Format | Length, structure, tone |
| 5. Example | A sample in the style you want |
| 6. Avoid | Phrases, assumptions or content to leave out |
Most weak AI answers come from vague prompts, not weak models. Giving the model what a capable new colleague would need fixes most of them.
prompt-builder/
├── SKILL.md # Workflow and rules Claude follows
├── references/
│ ├── framework.md # Each part in depth, safeguards, model notes
│ └── examples.md # Before/after prompts
├── scripts/
│ ├── check_prompt.py # Scores a prompt against the six parts
│ └── build_prompt.py # Assembles a prompt from JSON fields
└── assets/
└── template.md # Blank fill-in template
The scripts use only the Python standard library.
prompt-builder.zip from the latest release, or zip the prompt-builder folder yourself.See Anthropic's guide: Use Skills in Claude. Skills need code execution enabled.
Copy the folder into your personal or project skills directory:
git clone https://github.com/Hussain-Shahid-Javaid/claude-prompt-builder-skill.git
cp -r claude-prompt-builder-skill/prompt-builder ~/.claude/skills/ # all projects
# or
cp -r claude-prompt-builder-skill/prompt-builder .claude/skills/ # this project only
# Check a prompt
python3 prompt-builder/scripts/check_prompt.py my_prompt.txt
echo "write an email to my boss asking for a day off" | python3 prompt-builder/scripts/check_prompt.py
# Build a prompt from fields
python3 prompt-builder/scripts/build_prompt.py fields.json
Example check_prompt.py output:
Prompt check: 1/6 parts found (11 words)
[x] Task
[ ] Context
[ ] Audience and purpose
[ ] Format
[ ] Example
[ ] Avoid
Missing essentials: Context
Warnings:
- Very short prompt (11 words). Most good prompts need at least a task, context and format.
fields.json keys: task (required), context, audience, format, example, avoid, plus optional booleans ask_questions and flag_uncertainty.
The checker uses simple heuristics. It points at likely gaps; it doesn't replace reading your prompt.
The same framework is available as a free browser tool (no sign-up): AI Prompt Builder. The longer explanation is in How to write better AI prompts.
Issues and pull requests are welcome, especially new before/after examples in references/examples.md and improvements to the checker's heuristics.
MIT. Made by PromptPulse.