sf-ai-agentforce-persona by jaganpro/sf-skills
npx skills add https://github.com/jaganpro/sf-skills --skill sf-ai-agentforce-persona当用户需要定义代理个性而非实现细节时使用此技能:品牌到人设的转换、语气/声音设计、人设文档、示例对话优化,或为 Agent Builder / Agent Script 进行人设编码。
当工作涉及以下内容时,使用 sf-ai-agentforce-persona:
当用户进行以下操作时,请转交其他技能处理:
.agent DSL 行为 → sf-ai-agentscript询问或推断:
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
当用户提供以下内容时使用:
当用户提供一份已完成的人设文档并要求将其转换为以下内容时使用:
如果情况不明确,请提出一个澄清性问题:是设计一个新的人设,还是编码一个已有的人设?
设计循环是:输入 → 草稿 → 示例对话 → 优化 → 下载
有效的输入包括:
.agent 片段如果输入已包含足够的信息,则无需进行冗长的信息收集。
优先提取上下文而非提问。仅询问仍不清楚的内容,通常是:
所有问题都应可跳过。
围绕以下方面起草:
如果没有直接证据,则使用框架默认值或最接近的原型作为起点。
在首次展示时,显示:
这使差异显而易见。之后,除非用户另有要求,否则仅重新生成应用人设的版本。
将“更亲切”、“不那么正式”、“更简短”或“更具个性”等请求映射到特定的属性调整。
如果用户要求查看设置,则显示属性表并让他们直接调整数值。
进行针对性更改后:
将最终文档写入:
_local/generated/[agent-name]-persona.md使用:
当人设已存在且用户希望获得平台就绪的输出时使用此流程。
仅收集与编码相关的上下文:
将编码输出写入:
_local/generated/[agent-name]-persona-encoding.md使用:
此技能最多可生成三个 Markdown 文件:
默认路径:
_local/generated/[agent-name]-persona.md_local/generated/[agent-name]-persona-scorecard.md_local/generated/[agent-name]-persona-encoding.md评分是按需进行的,非自动执行。
50 分评分标准侧重于:
如果某个类别得分较低,请在编码前准确说明需要加强的内容。
| 需求 | 委托给 | 原因 |
|---|---|---|
| 构建主题/操作/元数据 | sf-ai-agentforce | 人设设计后的实现 |
将行为编码到 .agent 逻辑中 | sf-ai-agentscript | 确定性脚本编写 |
| 验证完成的代理行为 | sf-ai-agentforce-testing | 构建后的测试 |
| 分数 | 含义 |
|---|---|
| 45–50 | 可用于生产环境的人设 |
| 35–44 | 基础良好,编码前需优化 |
| 25–34 | 需要修改以提高一致性 |
| < 25 | 需要从身份和意图重新开始 |
每周安装次数
153
代码仓库
GitHub 星标数
223
首次出现
2026年3月4日
安全审计
安装于
cursor152
gemini-cli150
codex150
opencode150
github-copilot149
amp149
Use this skill when the user needs a defined agent personality , not implementation details: brand-to-persona translation, tone/voice design, persona documents, sample-dialog refinement, or persona encoding for Agent Builder / Agent Script.
Use sf-ai-agentforce-persona when the work involves:
Delegate elsewhere when the user is:
.agent DSL behavior → sf-ai-agentscriptAsk for or infer:
Use when the user provides:
Use when the user provides a completed persona document and asks to turn it into:
If ambiguous, ask a single clarifying question: design a new persona, or encode an existing one?
The design loop is: input → draft → sample dialog → refine → download
Valid inputs include:
.agent excerptDo not force a long intake if the input already contains enough signal.
Prefer extracting context before asking. Ask only for what is still unclear, typically:
All questions should be skippable.
Draft around:
If no direct evidence exists, use the framework defaults or nearest archetype as a starting point.
On the first reveal, show:
This makes the delta obvious. After that, regenerate only the persona version unless the user asks otherwise.
Map requests like “warmer”, “less formal”, “shorter”, or “more personality” to specific attribute shifts.
If the user asks to see settings, show the attribute table and let them adjust values directly.
After a targeted change:
Write the final document to:
_local/generated/[agent-name]-persona.mdUse:
Use this when a persona already exists and the user wants platform-ready output.
Gather only encoding-specific context:
Write the encoding output to:
_local/generated/[agent-name]-persona-encoding.mdUse:
This skill can produce up to three Markdown files:
Default paths:
_local/generated/[agent-name]-persona.md_local/generated/[agent-name]-persona-scorecard.md_local/generated/[agent-name]-persona-encoding.mdScoring is on-demand, not automatic.
The 50-point rubric focuses on:
If a category scores low, explain exactly what to strengthen before encoding.
| Need | Delegate to | Reason |
|---|---|---|
| build topics / actions / metadata | sf-ai-agentforce | implementation after persona design |
encode behavior into .agent logic | sf-ai-agentscript | deterministic script authoring |
| validate finished agent behavior | sf-ai-agentforce-testing | post-build testing |
| Score | Meaning |
|---|---|
| 45–50 | production-ready persona |
| 35–44 | strong foundation, refine before encoding |
| 25–34 | needs revision for coherence |
| < 25 | restart from identity and intent |
Weekly Installs
153
Repository
GitHub Stars
223
First Seen
Mar 4, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykWarn
Installed on
cursor152
gemini-cli150
codex150
opencode150
github-copilot149
amp149
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
63,800 周安装