Token导航 LogoToken导航

Agent Skill · 研究检索

prospect-refresh(前景刷新)

prospect-refresh 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 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。

prospect-refresh 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为调用外部服务;许可证信息还不明确,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 69,087。

总安装 220周安装 9GitHub Stars 12下载 71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:prospect-refresh(前景刷新)
来源仓库:https://github.com/scientiacapital/skills
仓库路径:skills/prospect-refresh
安装命令:
npx skills add https://github.com/scientiacapital/skills --skill prospect-refresh
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/scientiacapital/skills --skill prospect-refresh

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。

prospect-refresh 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。适用于需要根据关键词或任务场景进行信息搜索与筛选的研究检索场景。通过 npx skills add 命令从指定 GitHub 仓库安装并使用。

安装前建议确认权限范围、维护状态及是否涉及联网或文件操作。具体用法可参考原始 README 和 SKILL.md 文档进一步核验。

来源仓库内路径:skills/prospect-refresh

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/scientiacapital/skills --skill prospect-refres…;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

    它需要调用外部服务,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

prospect-refresh 怎么安装?

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

prospect-refresh 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。

prospect-refresh 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

prospect-refresh 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。

prospect-refresh 用的人多吗?

公开数据是累计安装 220 次,近一周安装 9 次,下载 71 次,GitHub 12 星。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。

prospect-refresh 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。

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。

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
prospect-refresh

前景刷新 · 本页

22012调用外部服务未明确有提醒

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

Prospect Refresh Skill

<quick_start> Trigger: Monday 6:30 AM ET (runs after prospect-enrich) Manual Trigger: "Run prospect refresh" or "Search new ICP prospects" Dependencies: Requires HubSpot portal 21530819, Apollo credits, Gmail tkipper@epiphan.com Output: 30 net-new prospects + 30 Gmail draft emails + HTML report with HubSpot links </quick_start>

<success_criteria>

  • Query Apollo for ICP title + vertical combinations (ATL-first targeting)
  • Deduplicate against HubSpot contacts (check email, company domain, name)
  • Filter Golden Rules (exclude customers, channel, device owners, etc.)
  • Enrich top 30: company revenue, headcount, funding, tech stack
  • Create Gmail draft for each prospect (personalized template per vertical)
  • Generate HTML report with sortable columns: name, title, company, vertical, ATL/BTL, links to HubSpot/Apollo
  • Include ATL/BTL tier badges and vertical ICP scores
  • Report: total prospects found, dedupe rate, ATL %, top 5 companies by headcount </success_criteria>

Stage 1: Define ICP Search Matrix

Verticals (7 ICPs) with targeting weights:

VerticalICP ScoreATL TitlesTarget CountBudget Authority
Higher Ed90CIO, VP IT, Dir. of IT Services, Provost, Dean, Dir. Academic Tech5>$50K
Courts85Clerk of Court, Court Administrator, Court Executive Director, Chief Judge4>$75K
Government80City Manager, IT Director/CIO, Director of Procurement, County Admin5>$100K
Corporate AV80VP Facilities, VP IT/CIO, VP Corp Communications, Dir. Facilities Ops, Dir. IT Infra5>$150K
Healthcare75CFO, CIO, COO, VP Operations, Director of Medical Education4>$100K
Houses of Worship70Senior Pastor, Executive Pastor, Finance Committee Chair, Facilities Chair4>$25K
K-1265Superintendent, CTO/Dir. Tech Services, Dir. Instructional Tech, Building Principal3>$10K

Total Target: 30 prospects (weighted by ICP score)


Stage 2: Query Apollo via People API Search

MCP Tool: apollo_mixed_people_api_search

Search Loop (iterate per vertical):

organization_locations: ["United States", "Canada"]
person_titles: [ATL_TITLES_FOR_VERTICAL]
organization_keyword_tags: [VERTICAL_KEYWORD] # e.g., "Higher Ed", "Education"
person_seniorities: ["c_suite", "vp", "director"]
organization_num_employees_ranges: [HEADCOUNT_RANGE_FOR_VERTICAL]
page: 1
per_page: 25

Example Query (Higher Ed):

person_titles: ["Chief Information Officer", "CIO", "VP Information Technology", "VP of IT", "Director of IT Services", "Director of Technology", "Provost"]
organization_keyword_tags: ["Higher Education", "Education", "University", "College"]
organization_num_employees_ranges: ["51,200", "201,500", "501,1000", "1001,5000", "5001,10000"]
person_locations: ["United States", "Canada"]

Output per vertical: Up to 25 results (name, email, title, organization, location, person_id, organization_id)


Stage 3: Deduplicate Against HubSpot

MCP Tool: search_crm_objects (HubSpot)

For each Apollo result, check:

filterGroups: [
  {
    filters: [
      { propertyName: "email", operator: "EQ", value: apollo_result.email }
    ]
  },
  {
    filters: [
      { propertyName: "company", operator: "CONTAINS_TOKEN", value: apollo_result.organization_name }
    ]
  }
]

Exclusion Logic:

  • If email exists in HubSpot → SKIP
  • If company + similar name exists → SKIP
  • If domain in blacklist (channel, device owners, etc.) → SKIP
  • Otherwise → KEEP

Expected Dedup Rate: 40-60% of Apollo results removed (existing contacts)


Stage 4: Apply Golden Rules + ATL/BTL Filter

Golden Rules (Hard Exclusions):

  • Skip if company in crm_customers (customer check)
  • Skip if contact tagged "Channel Partner"
  • Skip if job title contains BTL keywords: Technician, Support, Intern, Admin, Engineer (non-director), Operator
  • Skip if title in NEVER ATL list (Warehouse Manager, Network Admin, AV Tech, etc.)
  • Keep: USA/Canada only, public company or funded (exclude bootstraps <$1M)

ATL/BTL Classification (per CLAUDE.md):

ATL Tier (Always Prospect):

  • Chief (C-suite), VP, Director (IT/Tech/Facilities), Provost, Dean
  • Court Administrator, Clerk of Court, City Manager, Senior Pastor

GRAY Tier (Budget Authority >$25K Required):

  • Manager roles with Director+ reporting line
  • Department Chair (context-dependent)
  • Program Director (dept-level budget only)

BTL Tier (Skip for ATL-first targeting):

  • Technician, Specialist, Coordinator, Support, Administrator
  • Engineer (non-director), Operator, Instructor, Designer, Assistant
  • Help Desk, Intern, Volunteer

Output after filtering: Typically 50-60% of dedup results (15-18 prospects per vertical)


Stage 5: Enrich Top 30 Results

MCP Tool: apollo_organizations_enrich (for company data)

For each retained prospect:

domain: organization.domain (from Apollo result or extract from email)

Enrich fields:

  • annual_revenue, headcount, founded_year
  • latest_funding_amount, latest_funding_date
  • website, industry, keyword_tags
  • technologies (current tech stack)

MCP Tool: find-and-enrich-contacts-at-company (Clay fallback for company data)

If Apollo org enrichment incomplete:

companyIdentifier: prospect.domain
dataPoints: {
  companyDataPoints: [
    { type: "Latest Funding" },
    { type: "Tech Stack" },
    { type: "Company Customers" }
  ]
}

Stage 6: Create Gmail Drafts (One per Prospect)

MCP Tool: gmail_create_draft

Per-Vertical Email Templates (Personalized):

Template A — Higher Ed (ICP 90):

To: prospect.email
Subject: Epiphan + [UNIVERSITY_NAME]: Hybrid Learning Infrastructure for [SCHOOL_TYPE]

Body:
Hi [FIRST_NAME],

I was researching [UNIVERSITY_NAME]'s commitment to hybrid learning and asynchronous instruction—I noticed the shift in your academic technology strategy.

We work with schools like [REFERENCE_UNIV] to solve a specific challenge: instructor burnout from manual video streaming and recording.

[COMPANY_NAME] provides:
- Live lecture capture (no manual setup)
- Auto-recordings to your LMS
- Multi-platform distribution (Zoom, Teams, Canvas)

I'm not trying to sell—just want to understand if this is a priority this quarter for your department.

Are you the right person to discuss academic video infrastructure, or should I loop in [ROLE_IF_DIRECTOR]?

Best,
Tim
--
Epiphan Video

Template B — Courts (ICP 85):

To: prospect.email
Subject: [COURT_NAME] Remote Proceedings + Video Evidence Management

Body:
Hi [FIRST_NAME],

I was reading about [COURT_NAME]'s recent adoption of remote hearing technology. Given the volume of video evidence you're handling, I suspect you're managing multiple formats and platforms.

We partner with courts in [STATE] to centralize video evidence capture and courtroom live-streaming.

Curious: are you managing this manually today, or do you have a platform in place?

Would be worth a 15-min conversation if you're evaluating vendors this quarter.

Best,
Tim

Template C — Corporate AV (ICP 80):

To: prospect.email
Subject: [COMPANY] Boardroom Modernization — 4K Capture + Distribution

Body:
Hi [FIRST_NAME],

I noticed [COMPANY] is expanding your meeting rooms and conference infrastructure. Given your footprint, I'm guessing you're wrestling with:
- Inconsistent video quality across rooms
- Integration with Zoom/Teams/corporate network
- Archival and compliance

We've solved this for companies like [REFERENCE_CORP] (similar size/industry).

Worth a quick call to explore?

Best,
Tim

Template D — Healthcare (ICP 75):

To: prospect.email
Subject: [HOSPITAL_NAME] Medical Education + Video Evidence

Body:
Hi [FIRST_NAME],

I was researching [HOSPITAL_NAME]'s surgical training and medical education programs. I see you're ramping up simulation labs and remote mentoring.

Epiphan powers video capture and distribution for teaching hospitals doing exactly this.

Quick question: who owns your medical education video infrastructure today?

Best,
Tim

Template E — Houses of Worship (ICP 70):

To: prospect.email
Subject: [CHURCH_NAME] Livestream + Archive

Body:
Hi [FIRST_NAME],

I noticed [CHURCH_NAME] is offering both in-person and online services. Curious how you're handling the livestream and on-demand library today.

We work with churches your size to simplify setup and reach.

Might be worth exploring—no commitment.

Best,
Tim

Template F — K-12 (ICP 65):

To: prospect.email
Subject: [DISTRICT_NAME] Hybrid Learning + Board Meeting Recording

Body:
Hi [FIRST_NAME],

I was researching how [DISTRICT_NAME] is scaling hybrid learning post-pandemic. Guessing you're balancing live instruction, recordings, and board meeting transparency.

Epiphan helps districts like [REFERENCE_DISTRICT] simplify video capture across buildings.

Quick question: what's your current capture setup?

Best,
Tim

Draft Creation Logic:

  • Choose template based on prospect's vertical
  • Substitute [BRACKETS] with actual prospect/company data
  • Do NOT send—leave as draft for manual review
  • Tag draft with prospect's ATL/BTL tier + vertical ICP score

MCP Tool Parameters:

to: prospect.email
subject: [PERSONALIZED_PER_VERTICAL]
body: [TEMPLATE_WITH_SUBSTITUTIONS]
contentType: "text/plain"

Stage 7: Generate HTML Report

Output Format: Self-contained HTML file with:

Report Header:

  • Execution date/time
  • Total prospects found (Apollo)
  • Dedupe rate %
  • Final net-new count
  • ATL/BTL breakdown
  • Top 5 verticals by prospect count

Sortable Table (JavaScript):

NameTitleCompanyVerticalICPATL/BTLHeadcountRevenueGmail DraftHubSpot LinkApollo Link
Jane SmithCIOHigher Education UnivHigher Ed90ATL8,500$2.1B✓ Draft[link][link]
Bob JonesCourt AdminState Supreme CourtCourts85ATL450$120M✓ Draft[link][link]

Summary Section:

  • Top 5 companies by headcount
  • Top 5 companies by funding (if available)
  • Vertical breakdown (counts by ICP)
  • ATL % vs. BTL % in results
  • Recommended dial priority (ATL first, >$500M revenue second)

Links Format:

  • HubSpot link: https://app.hubspot.com/contacts/21530819/record/0-1/{contactId} (placeholder until contact created)
  • Apollo link: https://app.apollo.io/search?filters... (from Apollo result)

Stage 8: Load into HubSpot (Optional Next Step)

Note: This skill outputs drafts + report. Sequence Load Skill runs 45 min later (7:15 AM) to:

  • Create HubSpot contacts from top 30
  • Validate phone numbers
  • Add to Apollo outreach sequence

Emit Outcome Sidecar

As the final step, write to ~/.claude/skill-analytics/last-outcome-prospect-refresh.json:

{"ts":"[UTC ISO8601]","skill":"prospect-refresh","version":"1.0.0","variant":"default",
 "status":"[success|partial|error]","runtime_ms":[estimated ms from start],
 "metrics":{"prospects_found":[n],"drafts_created":[n],"atl_count":[n],"verticals_searched":[n]},
 "error":null,"session_id":"[YYYY-MM-DD]"}

Use status "partial" if some stages failed but results were produced. Use "error" only if no output was generated.


Skill Metadata

Version: 1.0 Last Updated: 2026-03-19 Author: Tim Kipper Status: Production Integration: Apollo + HubSpot (portal 21530819) + Gmail Tier: P1 (Core BDR Automation) Triggers: Scheduled (Monday 6:30 AM) + Manual ("Run prospect refresh") Dependencies: prospect-enrich (runs 6:00 AM), feeds into sequence-load (7:15 AM)

平台分布

Codex

36.97%
按下载量换算26

Claude

28.12%
按下载量换算20

Cursor

18.34%
按下载量换算13

Gemini CLI

8.26%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

可疑

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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