Token导航 LogoToken导航

Agent Skill · 开发

personize-signal(拟人化信号)

personize-signal 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

personize-signal 是开发类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描有提醒项,按安装量在开发类的 21,746 个 Skill 里排第 20,060。

总安装 242周安装 10GitHub Stars 公开资料未说明下载 79核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:personize-signal(拟人化信号)
来源仓库:https://github.com/personizeai/personize-skills
仓库路径:skills/personize-signal
安装命令:
npx skills add https://github.com/personizeai/personize-skills --skill personize-signal
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/personizeai/personize-skills --skill personize-signal

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

用于处理 GitHub 仓库、Issue、Pull Request 等协作信息。适合围绕代码变更、仓库状态进行整理和分析。通过 GitHub 安装,支持主流 AI 编程工具集成。使用前需确认权限范围和仓库维护状态。personize-signal 属于开发类 Skill,可作为该场景下的辅助能力补充。

建议结合原始 README 核验具体用法,避免触发不必要的命令执行。

来源仓库内路径:skills/personize-signal

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/personizeai/personize-skills --skill personize…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

personize-signal 怎么安装?

在终端执行 npx skills add https://github.com/personizeai/personize-skills --skill personize-signal。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

personize-signal 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

personize-signal 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

personize-signal 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

personize-signal 用的人多吗?

公开数据是累计安装 242 次,近一周安装 10 次,下载 79 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

personize-signal 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理开发类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
personize-signal

拟人化信号 · 本页

242—只读未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Skill: Personize Signal

Set up AI-powered notifications that decide IF, WHAT, WHEN, and HOW to notify each person. Signal uses Personize memory and governance to send the right message, to the right person, at the right time — or stay quiet when silence is better.

What This Skill Solves

Most notification systems are dumb rules: "if event X, send template Y." They don't know who the person is, what they care about, what was already sent, or whether this notification will help or annoy.

Signal replaces rules with intelligence:

  • IF — Should this person be notified at all? (AI scores 0-100 based on full entity context)
  • WHAT — What should the message say? (Uniquely personalized using everything known about the person)
  • WHEN — Now, or later in a digest? (AI decides priority: immediate / standard / digest)
  • HOW — Which channel? (Email, Slack, in-app, SMS — chosen based on context)

The intelligence comes from Personize SDK — memory, governance, and AI. Signal packages the orchestration into a reusable engine.


When This Skill is Activated

If the developer mentions notifications, alerts, or messaging, start with ASSESS.

If the developer has a specific question (e.g., "how do I add a Slack channel to Signal"), jump to the relevant action.

If the developer wants to understand the architecture, walk them through the engine flow.


When NOT to Use This Skill

  • Need raw memory operations → use entity-memory
  • Need shared workspaces without notifications → use collaboration
  • Need governance rules without notifications → use governance
  • Need CRM data sync → use entity-memory (CRM sync section)

Actions

ActionWhen to Use
ASSESSUnderstand their stack, events, channels, and recipients
CONFIGUREGenerate Signal initialization code
CONNECTScaffold event hooks for their framework
GOVERNSet up governance rules for notification guidelines
TESTRun a dry-run evaluation and verify the decision

Action: ASSESS

Understand what the developer needs before writing code.

Questions to Ask

  1. What events matter? — What happens in your product that should potentially trigger a notification?

- User lifecycle: signup, login, trial expiry, churn risk - Usage: milestones, drops, feature adoption - CRM: deal changes, meeting outcomes, support tickets - System: sync complete, errors, billing events

  1. Who are the recipients? — Who gets notified?

- End users (in-app, email) - Internal team (Slack, email) - Both (different channels)

  1. What channels? — How should messages be delivered?

- Email (SES, SendGrid) - Slack (webhook) - In-app (bridge to existing UI) - SMS (Twilio — community channel)

  1. What governance rules exist? — What constraints apply?

- Max notifications per day per user - Quiet hours / time zones - Content policies (tone, compliance) - Opt-out preferences

  1. What framework? — Where do events originate?

- Express.js / Fastify / Next.js / NestJS - Trigger.dev / n8n / cron jobs - Webhook-based (external system pushes events)

Output

After assessment, summarize:

  • Events to handle (with types like user.signup, usage.drop)
  • Channels needed
  • Governance constraints
  • Framework/integration approach

Action: CONFIGURE

Generate the Signal initialization code based on the assessment.

Minimal Setup

import { Personize } from '@personize/sdk';
import { Signal, ConsoleChannel, ManualSource } from '@personize/signal';

const client = new Personize({ secretKey: process.env.PERSONIZE_SECRET_KEY! });
const manual = new ManualSource();

const signal = new Signal({
    client,
    channels: [new ConsoleChannel()],
    sources: [manual],
});

await signal.start();

Production Setup

import { Personize } from '@personize/sdk';
import {
    Signal,
    ManualSource,
    SesChannel,       // or SendGridChannel
    SlackChannel,
    InAppChannel,
} from '@personize/signal';

const client = new Personize({ secretKey: process.env.PERSONIZE_SECRET_KEY! });
const manual = new ManualSource();

const signal = new Signal({
    client,
    channels: [
        new SesChannel({ sourceEmail: 'notifications@yourapp.com' }),
        new SlackChannel({ webhookUrl: process.env.SLACK_WEBHOOK! }),
        new InAppChannel(async (recipient, payload) => {
            // Bridge to your existing notification UI
            await yourNotificationService.create({
                userId: recipient.userId,
                title: payload.subject,
                body: payload.body,
            });
            return { success: true, channel: 'in-app', timestamp: new Date().toISOString() };
        }),
    ],
    sources: [manual],
    engine: {
        dailyCap: 5,                              // max per user per day
        deduplicationWindowMs: 6 * 60 * 60 * 1000, // 6 hours
        memorize: true,                            // feedback loop
        workspaceUpdates: true,                    // track in workspace
    },
});

await signal.start();

// Schedule daily digests (weekdays 9 AM)
signal.schedule('daily-digest', '0 9 * * 1-5', async () => {
    const users = await getActiveUsers();
    await signal.digest.runBatch(users);
});

Configuration Reference

OptionDefaultDescription
engine.dailyCap5Max notifications per email per day
engine.deduplicationWindowMs6hSkip same event type within window
engine.memorizetrueRecord sent notifications in memory
engine.workspaceUpdatesfalseCreate workspace entries on SEND/DEFER
engine.concurrency5Max parallel evaluations
engine.maxEvaluationsPerMinute20Rate limit for batch processing

Action: CONNECT

Scaffold event hooks that push product events into Signal.

Express.js Pattern

// event-hooks.ts — fire-and-forget, never throws, never blocks
import { ManualSource } from '@personize/signal';

const manual = new ManualSource(); // same instance used in Signal config

export const EventHooks = {
    async onUserSignup(email: string, data?: Record<string, unknown>) {
        manual.emit({
            id: `evt_${Date.now()}_${Math.random().toString(36).substring(2, 6)}`,
            type: 'user.signup',
            email,
            data: data || {},
            timestamp: new Date().toISOString(),
            metadata: { team: 'product' },
        });
    },

    async onUsageDrop(email: string, data: { metric: string; dropPercent: number }) {
        manual.emit({
            id: `evt_${Date.now()}_${Math.random().toString(36).substring(2, 6)}`,
            type: 'usage.drop',
            email,
            data,
            timestamp: new Date().toISOString(),
            metadata: { team: 'product' },
        });
    },
};

Using in Controllers

// In your signup controller
router.post('/signup', async (req, res) => {
    const user = await createUser(req.body);
    EventHooks.onUserSignup(user.email, { plan: user.plan }).catch(() => {});
    res.json(user);
});

Webhook Source (external events)

import { WebhookSource } from '@personize/signal';

const webhookSource = new WebhookSource({
    path: '/webhooks/signal',
    secret: process.env.WEBHOOK_SECRET,
    parser: (body) => ({
        id: body.id || `wh_${Date.now()}`,
        type: body.event || 'webhook.received',
        email: body.email || body.user_email,
        data: body.data || body,
        timestamp: body.timestamp || new Date().toISOString(),
    }),
});

// Mount on Express
app.use(webhookSource.middleware());

Action: GOVERN

Set up governance rules that guide Signal's AI decisions.

Governance Variables to Create

Use the Personize web app or SDK to create these governance variables:

1. Notification Guidelines (tag: notifications)

Notification Policy:
- Never notify about events the user has already seen in the product
- Prioritize actionable insights over informational updates
- Match tone to the urgency: critical = direct, informational = friendly
- Always include a specific next step, not generic "check it out"
- Maximum 3 notifications per user per day (engine enforces 5 as hard cap)

2. Channel Routing (tag: notifications)

Channel Selection:
- Slack: For internal team alerts (sales, support, ops)
- Email: For user-facing notifications that need persistence
- In-app: For real-time product notifications when user is active
- Digest: For low-priority updates that can wait for weekly summary

3. Frequency Policy (tag: communications)

Communication Frequency:
- New users (< 30 days): Max 1 notification per day. Focus on onboarding milestones.
- Active users: Max 3 per week. Only high-signal events.
- At-risk users: Increase frequency for re-engagement. Max 1 per day.
- Churned users: Do not notify. Only send win-back campaigns via marketing.

How Governance Flows into Signal

Signal's engine calls smartGuidelines() and smartDigest() in step 3 (context assembly). The AI sees these governance rules alongside the entity's full context and makes decisions accordingly. No code changes needed — update governance variables and Signal adapts.


Action: TEST

Verify Signal is working with a dry-run evaluation.

Quick Test

const result = await signal.trigger({
    id: 'test_001',
    type: 'user.signup',
    email: 'test@example.com',
    data: { plan: 'trial', source: 'website' },
    timestamp: new Date().toISOString(),
});

console.log('Action:', result.action);    // SEND | DEFER | SKIP
console.log('Score:', result.score);      // 0-100
console.log('Reasoning:', result.reasoning);
console.log('SDK calls:', result.sdkCallsUsed);
console.log('Duration:', result.durationMs, 'ms');

What to Verify

  1. Pre-check works — Trigger the same event twice within 6 hours. Second should SKIP instantly (0 SDK calls).
  2. Daily cap works — Trigger 6 events for the same email. The 6th should SKIP.
  3. AI decision varies — Trigger different event types. Scores should vary based on context relevance.
  4. Feedback loop — After a SEND, check that the notification was memorized: client.memory.recall({query: 'signal:sent', email: '...'}).
  5. Dedup via memory — After step 4, trigger a similar event. The AI should reference the recently sent notification in its reasoning.

Console Channel for Testing

Use ConsoleChannel during development — it logs decisions without delivering:

const signal = new Signal({
    client,
    channels: [new ConsoleChannel()],  // logs to stdout
    sources: [manual],
});

Available Resources

ResourceContents
../../signal/README.mdFull Signal documentation — architecture, channels, sources, workspace, digest, cost controls
../../signal/CLAUDE.mdAI tool instructions for Signal package
../../signal/examples/quickstart/Minimal setup example
../../signal/examples/saas-onboarding/Signup → nurture → convert sequence
../../signal/examples/sales-alerts/Usage drop alerts to sales team
../../signal/examples/weekly-digest/Deferred items → compiled digest
../../signal/examples/multi-team/Product + Sales + Marketing on same records
../../signal/templates/CHANNEL_TEMPLATE.mdHow to build a new channel
../../signal/templates/SOURCE_TEMPLATE.mdHow to build a new source
recipes/quickstart.tsMinimal Signal setup recipe
recipes/multi-team.tsMulti-team configuration recipe

平台分布

Codex

39%
按下载量换算31

Claude

28.59%
按下载量换算23

Cursor

19.1%
按下载量换算15

Gemini CLI

9.98%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills