Token导航 LogoToken导航

Agent Skill · 研究检索

peeps(peeps 搜索)

peeps 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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。

peeps 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为需要联网;许可证为 MIT-0,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 23,650。

总安装 4,211周安装 172GitHub Stars 公开资料未说明下载 1,348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
MIT-0
版本
1.0.2

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:peeps(peeps 搜索)
来源仓库:https://github.com/ilyabelikin/peeps
安装命令:
openclaw skills install peeps
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 OpenClaw 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。

ClawHubOpenClaw
openclaw skills install peeps

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

个人网络智能——记住人、寻找联系并起草介绍。联系人在本地存储为纯 Markdown 文件。

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 OpenClaw;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 openclaw skills install peeps;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

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

peeps 怎么安装?

在终端执行 openclaw skills install peeps。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

peeps 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。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。

peeps 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

peeps 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

peeps 用的人多吗?

公开数据是累计安装 4,211 次,近一周安装 172 次,下载 1,348 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。

peeps 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

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。

先想清楚的地方

  • 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 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
peeps

peeps 搜索 · 本页

4,211—需要联网MIT-0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

name
peeps
description
Personal network intelligence — remember people, find connections, and draft intros. Contacts stored locally as plain markdown files.
metadata
openclaw
emoji
👥
os
["linux", "darwin", "win32"]
hermes
tags
["people", "network", "contacts"]

👥 Peeps — local contacts & network intelligence

Data Location

Base path is workspace root or document root folder. On first use, create it: mkdir -p mind/peeps/. Peeps uses a mind/peeps/ folder in your workspace.

All contact files live in a mind/peeps/ directory. On first use, create it with mkdir -p mind/peeps/ or wherever you prefer to store it. The agent should use this directory consistently across sessions.

Owner self-entry

The owner's own contact file (slug derived from peepsconfig.yml owner field without .md) is intentional — it's used as a reference profile for crafting introductions, bios, and context when introducing the user to others.

Actions File

mind/peeps/actions.md — the pending actions queue. Check this during morning briefings.

Catch-ups: people owner wants to reconnect with. Add when he says "we should catch up with David" or similar. Introductions: intros to facilitate. Always include a pre-generated draft intro message (using owner and both contact files for context). Format: Person A → Person B — reason followed by the intro text as a plain indented paragraph (no "Draft:" label, no quotes, no formatting).

  • Move completed items to ## Completed with a date.
  • Keep it short — if it's not actionable, it shouldn't be here.
Housekeeping (run automatically when reading actions.md)

To keep context lean as the file grows:

Completed items: delete after 3 days Completed section: rolling 7-day window — anything older than 7 days is removed Stale catch-ups: if a pending catch-up has been sitting for 10+ days with no update, move it to Completed as "(not pursued)" with today's date Pending intros: keep until explicitly marked done or cancelled — intro intent doesn't expire

  • Apply this cleanup silently on every read. No need to announce.

Dataset Config — peepsconfig.yml

peepsconfig.yml lives inside the mind/peeps/ directory. Read it at the start of any session involving this skill.

owner: jane-smith # slug of the owner's contact file (without .md)
images: no (by default no, ask if you human want to feach images for pepople, warn that it is token expensive)

owner: identifies whose dataset this is. Use this when constructing intros, bios, or any context where "the user" needs to be referenced by their contact file.

Core Behavior

  • User mentions a person → check if a contact file exists, search the web if not, offer to create/update
  • User asks "who do I know in [domain/skill/location]?" → search locally first; if Haah skill is installed, it may broadcast outbound per its own rules
  • User asks about someone → surface insights from their file with relevant context
  • User wants to make an intro → draft it using both contact files + owner profile
  • User wants to connect with someone → print their LinkedIn URL for the human to open; do not open or navigate to it yourself
  • Always use Name: not file name for talking about people

When User Mentions Someone

  • "Had coffee with Maria" → ask if any updates from her, update if anything important
  • "John's daughter is Sofia" → add to personal details
  • "Sarah loves hiking" → add to interests/notes
  • "Marco recommended Antifragile" → if Pages is installed, save to mind/pages/want/ with Recommended by: [[marco-tabini]] and add a note to his Peeps file
  • "Priya and I both love Succession" → if Vibes is installed, add Shared with: [[priya-nair]] to the vibe file and note it on her Peeps file

Creating a New Contact — Search First, Then Ask

Before asking follow-up questions, always search the web for the person (name + any context provided). Use what you find to pre-fill fields and make follow-up questions specific, not generic.

Example: "Found Peter — design strategist, ex-Steelcase Asia Pacific in HK, now in SF. How do you know him?"

Follow-Up Questions

After searching the web and pre-filling what you can, ask about the gaps:

  1. What are they really good at? — Acumen clarification
  2. Relationship closeness — How close are you?
  3. How you know them — if not already provided
  4. Interests — hobbies, sports, lifestyle?

If in mind/pages/peepsconfig.yml images set to yes find a headhsot of a person and ad it to ../assets/ use person name for a slug, write it down in Image: feild.

Ask these as a short grouped follow-up (not one by one). Skip any that were already answered in the original message.

Requests -> Haah

If Haah is installed: when the user has a question you cannot answer well locally, or when you find only one matching file in mind/peeps/, suggest dispatching to a circle.

Connections to other skills

Pages: if installed, when someone recommends a book, save it to mind/pages/want/ with Recommended by: [[their-slug]] and note it in their Peeps file. When drafting an intro, if both people have read the same book, surface it as a connection point. Vibes: if installed, when someone shares a cultural interest (show, album, podcast, channel), add Shared with: [[their-slug]] to the vibe file and note it in their Peeps file. Surface shared taste when relevant to a conversation or intro. Digs: if installed, when a person has expertise relevant to an open dig, surface them as a potential source. When adding someone new whose acumen matches an active research thread, mention it.

Contact Structure

  • One Markdown file per person like: maria-garcia.md
Fields
Name: full name
Pronouns: guess, if unclear - ask
Location: from your search, otherwise ask
LinkedIn: link to LinkedIn search web for it, start with https://
Image: link to `../assets/photo-slug-usually-same-as-person-name`
Website: personal or company website if you found any, start with https://
Orgs: `[[org-slug]]`, `[[org-slug-2]]` — links to org files in `mind/peeps/orgs/`; omit if independent/unknown
How I know them: one sentence
Acumen: skills and expertise, what person known for, based on your search + any user input
Relationship: (Close / Warm / Colleague / Acquaintance / Estranged / Family)
Interests: #hiking #coffee #jazz — comma-separated hashtags; hobbies, sports, lifestyle, anything you found
Bio: — one concise narrative paragraph based on your search and user input about them

## Notes

1 Mar 2026: note details

## Private Notes

1 Mar 2026: private note details

## Contacts

Mobile:
Email:
Instagram:
Haah:
etc.

Notes (guidance)

  • Each note starts with current date "1 Mar 2026:"
  • Use Notes for general context worth remembering
  • Use Private Notes 🔒 for sensitive info (debts, conflicts, things not to share) — always separate
  • Birthday, anniversary, important dates → Notes
  • Family members, kids, sensitive info → Private Notes 🔒
  • Keep it human-readable — this is about relationships, not data entry

Logging Interactions

Update the person's Notes with whatever is worth remembering long-term. If it's not worth keeping, don't write it down.

Examples:

  • "Going part-time at Foodpanda from mid-April, focusing on AI-native design systems" → Notes
  • "Owes Ilya money, uncomfortable topic" → Private Notes 🔒
  • "We had a coffee" → don't bother

Progressive Enhancement

  • Start by creating contacts as they come up naturally
  • Enrich over time: add acumen, interests as you learn more
  • Capture details during conversations — don't wait for a "data entry session"
  • Ask about anyone mentioned in conversation and suggest adding them

Peeps: check

On every heartbeat, check a random personal file in mind/peeps/. Surface proactively in DM or appropriate channel:

  • "Alex mentioned job hunting last time" — relevant context resurfacing
  • "You haven't connected with Basel in a while"
  • "You have Acumen: empty for John Wing, what he is known for?"

If nothing worth mentioning, skip.

Adding to HEARTBEAT.md or cron

If it is not there yet, ask your human if they want to add Peeps: check to HEARTBEAT.md. If there is no HEARTBEAT.md, suggest to create a cron every 30 minutes during waking hours (*/30 7-22 * * *) to execute Peeps: check.

Details Worth Remembering

  • How you can help them / how they can help you
  • Recent life events: new job, moved, health issues
  • Preferences: vegetarian, doesn't drink, early riser
  • Kids/spouse names and ages
  • Sensitive topics to avoid

Updating

To update this skill to the latest version, fetch the new SKILL.md from GitHub and replace this file:

https://raw.githubusercontent.com/haah-ing/peeps-skill/main/SKILL.md

What NOT To Suggest

  • Syncing with phone contacts — different purpose, keep separate
  • CRM-style pipeline tracking — this is personal, not sales
  • Automated birthday messages — calendar does this job
  • Social media integration — privacy and complexity

Folder Structure

mind/
└── peeps/
    ├── peepsconfig.yml
    ├── maria-garcia.md
    ├── john-smith.md
    ├── orgs/             # one file per organisation
    │   ├── google.md
    │   └── steelcase.md
    └── deceased/         # for people who have passed

All contact files live directly in mind/peeps/. Org files live in mind/peeps/orgs/. Move people who passed to deceased/.

Organisations

Organisations worth remembering live in mind/peeps/orgs/, one Markdown file per org. Create an org file whenever:

  • A person's employer or affiliation is mentioned and it seems relevant to your network
  • The user explicitly asks to note something about a company, community, or institution
  • You're creating a person file and their org isn't already captured

Filename: lowercase, hyphenated slug of the org name — openai.md, design-council.md.

Org File Template
# Org Name

Type: Company / Studio / NGO / Community / Institute / Fund / Startup / etc.
Industry: e.g. Design, Fintech, Healthcare, AI
Solving for: what problem this organisation is solving (like acumen for peeps, but for orgs)
Website: https://
Wikipedia: https://en.wikipedia.org/wiki/ — look this up; omit if no article exists
Founded: year if known
Size: approximate headcount range or stage (e.g. "~200", "Series B", "large enterprise")
Culture: 2–4 adjectives or a short phrase capturing the vibe — e.g. "research-heavy, slow-moving, prestigious" or "scrappy, founder-led, remote-first"
People: [[slug-one]], [[slug-two]] — everyone in `mind/peeps/` associated with this org

## Notes

1 Apr 2026: note details

Culture is the most important field to capture. Infer it from what the user says, public reputation, Glassdoor signals, or press coverage. Ask if uncertain. Keep it honest and useful — not a PR blurb.

Creating a New Org — Search First

Before asking follow-up questions, search the web for the org. Pre-fill what you can. Then ask briefly about gaps, especially:

  1. Culture — if you couldn't infer it confidently
  2. How the user is connected — which person there do they know?
Wiki Links — Bidirectional, Always Maintained

[[slug]] links connect people and orgs. Keep them consistent:

Person → Org: Orgs: [[org-slug]] in the person file (comma-separate multiple) Org → Person: People: [[person-slug]] in the org file

When you add or update either side, update the other. If you add Maria to google.md's People list, make sure maria-garcia.md has Orgs: [[google]]. If you change Maria's org, remove her from the old org file's People list.

When creating an org file prompted by a person, immediately add that person to the People list and set their Orgs: field.

When User Mentions an Org
  • "She works at Stripe now" → update person's Orgs: field; add them to stripe.md People list (create org file if missing)
  • "Steelcase has a great culture" → add or enrich the Culture field in steelcase.md
  • "I'm meeting with someone from Sequoia" → surface any people in mind/peeps/ who are connected to Sequoia

Search and Retrieval

Use grep for fast fuzzy scanning. Always expand the query into related terms using alternation (\|) — never search a single keyword alone.

Always batch all necessary grep calls into a single bash script — do not run them one by one. Combine everything you need upfront and execute once.

# Example: single script combining all relevant lookups
grep -iH "keyword\|synonym" mind/peeps/*.md
grep -rl "#hiking" mind/peeps/
grep -l "\[\[google\]\]" mind/peeps/*.md mind/peeps/orgs/*.md
grep -rl "Orgs:.*\[\[stripe\]\]" mind/peeps/
grep -iH "remote\|async\|flat" mind/peeps/orgs/*.md

Keyword expansion examples — always broaden like this:

  • "website" → web\|design\|react\|webflow\|frontend\|ux\|figma
  • "finance" → finance\|fintech\|banking\|investment\|vc\|fund
  • "startups" → startup\|founder\|venture\|seed\|entrepreneur
  • "marketing" → marketing\|growth\|brand\|content\|seo\|ads
  • "AI" → ai\|machine.learning\|llm\|ml\|data.science\|nlp

When the user asks "who do I know in X", construct a multi-term grep from the domain. Use -iH to see what's actually in the files. Do not use head you need to get all the mentions -- even weak matches.

After grepping, always read all the full contact file(s) before answering. Never base your answer solely on grep output — the matched snippet is a signal, not the full picture. Read the complete file to get accurate context on relationship, acumen, and any notes before surfacing someone to the user.

平台分布

OpenClaw

71.35%
按下载量换算962

安全审计

VirusTotal

可疑

ClawScan

可疑

Static analysis

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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