Token导航 LogoToken导航

Agent Skill · 研究检索

unloopa-api(unloopa API 搜索)

用于辅助 API 设计、接口文档、请求响应结构和服务集成说明。它适合让 Agent 梳理 endpoint、生成 OpenAPI 草稿、检查字段命名、整理错误码或辅助前后端联调。使用时需要确认真实业务语义、鉴权方式、分页和错误处理规则;涉及生成接口文档时,应避免凭空补字段,最好从现有代码、schema 或接口样例中提取事实。

unloopa-api 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 4,080。

总安装 36,338周安装 1,545GitHub Stars 4下载 12,731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
1.3.0

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:unloopa-api(unloopa API 搜索)
来源仓库:https://github.com/echris6/unloopa-api
安装命令:
openclaw skills install unloopa-api
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 unloopa-api

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。

让您的代理商自动向当地企业销售网站。从 Google 地图中查找潜在客户,为每个潜在客户构建一个自定义 AI 网站,发送外展电子邮件,甚至可以给他们打电话。当用户想要查找潜在客户、生成网站、发送电子邮件或拨打语音电话时使用。

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 unloopa-api;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

    它需要接触敏感数据,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

unloopa-api 怎么安装?

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

unloopa-api 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。

unloopa-api 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

unloopa-api 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。

unloopa-api 用的人多吗?

公开数据是累计安装 36,338 次,近一周安装 1,545 次,下载 12,731 次,GitHub 4 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。

unloopa-api 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

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。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 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。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 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
unloopa-api

unloopa API 搜索 · 本页

36,3384接触敏感数据MIT-0通过

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
unloopa-api
description
Make your agent sell websites to local businesses on autopilot. Finds leads from Google Maps, builds a custom AI website for each one, sends outreach emails, and can even call them. Use when the user wants to find leads, generate websites, send emails, or make voice calls.
metadata
author
unloopa
version
1.3
openclaw
primaryEnv
UNLOOPA_API_KEY
requires
env

Unloopa API

You control the Unloopa platform through its REST API. All requests go to https://dashboard.unloopa.com/api/v1/ with Bearer token authentication.

Authentication

Every request needs the header:

Authorization: Bearer $UNLOOPA_API_KEY

The API key is set in the UNLOOPA_API_KEY environment variable. Keys start with unlpa_live_.

If the user hasn't configured their key yet, tell them:

  1. Go to https://dashboard.unloopa.com/settings and click the "API" tab
  2. Click "Create API Key" and copy the key (it's only shown once)
  3. Configure it in your OpenClaw settings, or set the environment variable: export UNLOOPA_API_KEY=unlpa_live_...

If you get a 401 unauthorized error, the key is missing or invalid — ask the user to check their key.

First Call: Always Start With GET /quota

Before doing anything else, call GET /quota to discover:

  • What plan the user is on (starter or pro)
  • Whether voice and video features are enabled
  • How many leads and credits remain
  • Purchase links if they need to upgrade or buy credits

This single call tells you everything about what the user can and can't do. Adapt your behavior based on the response:

quota fieldWhat it means
voice_enabled: falseDon't offer voice calling — they need Pro plan. Share purchase_links.upgrade
video_enabled: falseDon't offer video generation — they need Pro plan
voice_credits: 0Can't make calls — share purchase_links.voice_credits
websites.remaining: 0Can't generate leads — quota resets at resets_at

Error Format

All errors return:

{ "error": { "code": "error_code", "message": "Human-readable message", "details": {} } }

Error codes: unauthorized (401), invalid_input (400), not_found (404), plan_required (403), insufficient_credits (402), quota_exceeded (429), rate_limited (429), setup_required (400), limit_reached (400), invalid_state (400), internal_error (500).

When you get plan_required (403), share the upgrade link from quota. When you get insufficient_credits (402), share the credit purchase links. When rate limited, check the Retry-After header (seconds).

Plans

  • Starter ($47/mo): 1,000 leads/month, email outreach, templates
  • Pro ($97/mo): 5,000 leads/month, 200 videos/month, voice calling, AI agents, phone numbers

Workflows

1. Full Lead Pipeline (any plan)

The /command endpoint runs the entire pipeline automatically: scrape leads → generate websites → enrich emails → send outreach. Just describe what you want.

1. GET  /quota                    → check websites.remaining > 0
2. POST /command                  → submit natural language command (full pipeline runs automatically)
3. GET  /jobs/{job_id}            → poll every 5-10s until status=completed
4. GET  /leads?job_id={job_id}    → view generated leads with websites, emails, etc.

2. Email Outreach (any plan)

1. GET  /outreach/status          → verify configured=true, remaining_today > 0
2. GET  /leads?has_email=true     → find leads with emails
3. GET  /outreach/templates       → pick a template
4. POST /outreach/send            → queue emails

If configured=false, tell the user to connect an email account at the setup_url in the response.

3. Voice Calling (Pro plan only)

Skip this workflow entirely if voice_enabled=false in /quota. Tell the user they need Pro and share the upgrade link.

Prerequisites: voice_enabled=true + voice_credits > 0 + at least 1 phone number + at least 1 voice agent.

1. GET  /quota                    → voice_enabled? voice_credits > 0?
   If voice_credits=0 → share purchase_links.voice_credits
2. GET  /phone-numbers            → count > 0? (max 3)
   If empty → POST /phone-numbers/search + POST /phone-numbers/buy
3. GET  /voice/agents             → count > 0? (max 3)
   If empty → POST /voice/agents (create one)
4. POST /voice/call               → single call, OR:
5. POST /voice/campaigns          → bulk campaign (starts as draft)
6. PATCH /voice/campaigns/{id}    → action=activate, then action=trigger

4. Full Funnel

The /command endpoint now handles steps 1-3 automatically. Voice calling is the only manual step.

1. GET  /quota                    → know the plan, adapt accordingly
2. POST /command → poll /jobs/{id} → GET /leads  (scrape + websites + emails + outreach all automatic)
3. Voice (Pro only): /phone-numbers → /voice/agents → /voice/campaigns

Endpoints Reference

POST /command

Submit a natural language lead generation command. The API automatically runs the full pipeline: scrape → generate websites → enrich emails/socials → send outreach. No need to mention each step in the command.

Body:

{
  "command": "Find 50 plumbers in Miami",
  "max_results": 50,
  "with_video": false,
  "with_vsl": false
}
  • command (required, string, max 1000 chars) — just describe the niche and location. Any number mentioned in the command is ignored — use max_results to control lead count.
  • max_results (optional, 1-100, default: 100, or 10 when with_video/with_vsl is true)
  • with_video (optional, bool, Pro plan only)
  • with_vsl (optional, bool, Pro plan only)

Default behavior: The API always overrides what's in the command text. It scrapes up to max_results leads (default 100), generates a website for each, finds email addresses, enriches social profiles, and sends outreach emails — all automatically. Numbers in the command like "Find 15 plumbers" are ignored; use max_results instead.

Response: { job_id, status: "processing", defaults: { max_results, generate_websites, enrich_emails, send_outreach, with_video, with_vsl }, quota: { used, limit, remaining } }


GET /jobs

List submitted commands.

Query: ?limit=20&offset=0 (limit max 100)

Response: { jobs: [{ job_id, command, intent, status, error, created_at, updated_at }], total, limit, offset }


GET /jobs/{id}

Poll a job for progress and results.

Response:

{
  "job_id": "uuid",
  "status": "processing|completed|failed",
  "progress": 75,
  "current_step": "Generating websites...",
  "steps": [{ "name": "scraping", "status": "completed", "message": "Found 50 leads", "count": 50 }],
  "result": {
    "websites": [{ "id": "uuid", "url": "https://...", "business_name": "...", "city": "...", "industry": "..." }],
    "leads_scraped": 50,
    "emails_sent": 0
  },
  "error": null
}

Poll every 5-10 seconds. Jobs take 30s to 5 minutes depending on count and video.


GET /leads

List and filter leads.

Query params (all optional):

  • limit (1-100, default 50), offset (default 0)
  • city — partial match (e.g. "miami")
  • industry — partial match (e.g. "plumber")
  • has_phone=true — only leads with phone numbers
  • has_email=true — only leads with email addresses
  • min_rating — minimum Google rating (e.g. 4.0)
  • min_reviews — minimum review count
  • job_id — leads from a specific command
  • search — free text search across name, city, industry
  • created_after — ISO date (e.g. "2025-01-15")
  • created_before — ISO date
  • has_website=true — only leads with generated website URLs
  • has_video=true — only leads with video
  • video_status — pending|generating|completed|failed

Response:

{
  "leads": [{
    "id": "uuid",
    "business_name": "Acme Plumbing",
    "city": "Miami",
    "industry": "Plumber",
    "phone": "+13055551234",
    "email": "info@acme.com",
    "rating": 4.8,
    "reviews": 127,
    "url": "https://unlora.com/acme-plumbing-miami",
    "language": "en",
    "video_url": "https://...",
    "video_status": "completed",
    "vsl_url": "https://...",
    "vsl_status": "completed",
    "socials": { "instagram": "...", "facebook": "...", "linkedin": "...", "twitter": "..." },
    "created_at": "2025-01-15T..."
  }],
  "total": 50, "limit": 50, "offset": 0
}

GET /leads/{id}

Full lead detail including existing website analysis.

Response: Same fields as list plus:

  • slug — URL slug
  • existing_website: { url, pagespeed_score, load_time, mobile_optimized } or null

GET /websites

Simpler list of generated websites.

Query: ?limit=20&offset=0

Response: { websites: [{ id, url, slug, business_name, city, industry, phone, email, language, video_url, vsl_url, created_at }], total, limit, offset }


GET /quota

Check plan, usage, credits, and purchase links.

Response:

{
  "plan": "pro",
  "plan_status": "active",
  "websites": { "used": 150, "limit": 5000, "remaining": 4850 },
  "videos": { "used": 0, "limit": 200, "remaining": 200 },
  "voice_credits": 45,
  "voice_enabled": true,
  "video_enabled": true,
  "resets_at": "2025-02-01T00:00:00.000Z",
  "purchase_links": {
    "voice_credits": {
      "50_credits_$10": "https://whop.com/checkout/plan_xBEWrVWZ8MRvM/",
      "200_credits_$35": "https://whop.com/checkout/plan_ucYBrssGb4E2G/",
      "500_credits_$75": "https://whop.com/checkout/plan_zTX2bQyWLCqlx/"
    },
    "upgrade": "https://whop.com/unloopa/"
  }
}

GET /outreach/status

Check email configuration, daily capacity, DNS health.

Response:

{
  "configured": true,
  "accounts": [{
    "id": "uuid",
    "email": "outreach@company.com",
    "display_name": "Company",
    "daily_limit": 25,
    "sent_today": 10,
    "remaining_today": 15,
    "warmup_enabled": true,
    "warmup_day": 14,
    "health": { "score": 85, "status": "good", "checked_at": "2025-01-15T..." },
    "created_at": "2025-01-01T..."
  }],
  "summary": {
    "total_accounts": 1,
    "total_daily_capacity": 25,
    "total_sent_today": 10,
    "total_remaining_today": 15,
    "pending_in_queue": 5,
    "accounts_with_health_issues": 0
  },
  "setup_url": "https://dashboard.unloopa.com/settings?tab=email"
}

New SMTP accounts warm up over 4 weeks: 5/day -> 10/day -> 15/day -> 25/day.


GET /outreach/templates

List prebuilt and custom email templates.

Response: { templates: [{ id, name, subject, body, is_custom: false, is_default: true, language }], custom_templates: [{ id, name, subject, body, is_custom: true, is_default, language }] }

Templates support placeholders: {{business_name}}, {{city}}, {{industry}}, {{website_url}}, {{video_url}} (Pro only).


POST /outreach/templates

Create custom email template.

Body:

{
  "name": "Miami Pitch",
  "subject": "{{business_name}} - New Website Ready",
  "body": "Hi! I built a website for {{business_name}} in {{city}}...",
  "language": "en",
  "is_default": true
}

Required: name, subject, body.


PATCH /outreach/templates/{id}

Update a custom template. Only custom templates can be edited.

Body: { name?, subject?, body?, language?, is_default? }


DELETE /outreach/templates/{id}

Delete a custom template.


POST /outreach/send

Send emails to leads.

Body:

{
  "lead_ids": ["uuid1", "uuid2"],
  "template_id": "uuid",
  "custom_subject": "Optional override",
  "custom_body": "Optional override"
}
  • lead_ids (required, 1-100 UUIDs)
  • template_id (optional if custom_subject + custom_body provided)

Response: { emails_queued, emails_waiting_for_video, skipped_duplicates, failed, manual_outreach: [] }

Requires SMTP configured (check /outreach/status first). Duplicate detection prevents re-sending.


GET /phone-numbers

List active phone numbers. Pro plan required.

Response: { numbers: [{ id, phone_number, area_code, locality, region, country, monthly_cost_cents, created_at }], count: 2, limit: 3 }

Max 3 phone numbers. Check count vs limit before buying.


POST /phone-numbers/search

Search available numbers by area code. Pro plan required.

Body: { "area_code": "305", "country": "US" }

  • area_code (required, 3 digits)
  • country (optional, default "US")

Response: { numbers: [{ phone_number: "+13055551234", friendly_name, locality, region }] }


POST /phone-numbers/buy

Purchase a phone number. Pro plan required. $1/month per number.

Body: { "phone_number": "+13055551234" } Must be E.164 format from search results.

Response: { number: { id, phone_number, area_code, monthly_cost_cents, created_at } }


DELETE /phone-numbers/{id}

Release a phone number. Pro plan required.

Response: { success: true }


GET /voice/agents

List voice agents. Pro plan required.

Response: { agents: [{ id, name, voice_id, voice_name, elevenlabs_agent_id, has_script, has_first_message, created_at }], count: 1, limit: 3 }

Max 3 agents.


POST /voice/agents

Create a voice agent. Pro plan required.

Body:

{
  "name": "Sales Agent",
  "voice_id": "cjVigY5qzO86Huf0OWal",
  "voice_name": "Eric",
  "script": "You are a friendly sales rep calling {{business_name}} in {{city}}...",
  "first_message": "Hi there, do you have just a moment?",
  "agent_config": { "stability": 0.3, "similarityBoost": 0.85 }
}

Required: name, voice_id, script.

Scripts support dynamic variables: {{business_name}}, {{city}}, {{industry}}, {{website_url}} — auto-populated from lead data during calls.


PATCH /voice/agents/{id}

Update a voice agent. Syncs changes to ElevenLabs automatically.

Body: { name?, voice_id?, voice_name?, script?, first_message?, agent_config? }


DELETE /voice/agents/{id}

Delete a voice agent. Also removes from ElevenLabs.


POST /voice/call

Make a single outbound call. Costs 1 voice credit. Pro plan required.

Body:

{
  "agent_id": "uuid",
  "phone_number": "+13055551234",
  "dynamic_variables": { "business_name": "Acme", "city": "Miami", "industry": "Plumbing", "website_url": "https://..." }
}

Required: agent_id, phone_number.

Response: { call_id, conversation_id, status: "initiated", phone_number }


GET /voice/calls

List voice calls with filters. Pro plan required.

Query: ?limit=50&offset=0&campaign_id=uuid&status=completed&outcome=interested

  • status: pending, queued, in_progress, completed, failed, cancelled
  • outcome: interested, not_interested, voicemail, no_answer, callback

Response: { calls: [{ id, campaign_id, business_name, phone_number, status, outcome, outcome_notes, duration_secs, transcript, analysis, started_at, completed_at, created_at }], total, limit, offset }


GET /voice/calls/{id}

Full call detail with transcript and analysis.

Response: { call_id, business_name, phone_number, status, outcome, outcome_notes, duration_secs, transcript, analysis, started_at, completed_at, created_at }


GET /voice/campaigns

List voice campaigns with stats. Pro plan required.

Response:

{
  "campaigns": [{
    "id": "uuid",
    "name": "Miami Plumbers",
    "status": "active",
    "stats": { "total": 50, "connected": 30, "interested": 8, "not_interested": 15, "no_answer": 7, "failed": 0, "avg_duration_secs": 45 },
    "created_at": "...", "updated_at": "..."
  }]
}

POST /voice/campaigns

Create a calling campaign. Starts as draft. Pro plan + voice credits required.

Body:

{
  "name": "Miami Plumbers Campaign",
  "phone_number_id": "uuid",
  "agent_id": "uuid",
  "lead_filter": { "city": "Miami", "industry": "plumber" },
  "timezone": "America/New_York",
  "calling_window_start": "09:00",
  "calling_window_end": "17:00",
  "calling_days": ["mon", "tue", "wed", "thu", "fri"],
  "calls_per_hour": 10,
  "max_calls": 50
}

Required: name, phone_number_id, and either agent_id OR voice_id + script. Lead selection: provide lead_ids (array of UUIDs) or lead_filter (dynamic). Only leads with phone numbers are included.

Response: { campaign: { id, name, status: "draft", leads_count, callable_leads, created_at } }


GET /voice/campaigns/{id}

Campaign detail with full config and stats.

Response includes: id, name, status, script, script_version, first_message, voice_id, voice_name, timezone, calling_window, calling_days, calls_per_hour, max_calls, stats (with pending count), timestamps.


PATCH /voice/campaigns/{id}

Control campaign lifecycle or update fields.

Lifecycle actions (body: { "action": "..." }):

  • activate — draft/paused -> active
  • pause — active -> paused
  • cancel — any -> completed (cancels pending calls)
  • trigger — active campaign: initiate up to 10 pending calls immediately. Each call costs 1 voice credit. Optional: { "action": "trigger", "limit": 5 }

Field updates (draft/paused only, body: { "updates": {...} }):

  • script, voice_id, voice_name, first_message, calling_window_start, calling_window_end, timezone, calling_days, calls_per_hour, agent_config

Trigger response: { triggered: 5, calls: [{ id, business_name, conversation_id }] }


Important Notes

  • Job processing takes ~8-10 minutes for a full pipeline run. Scraping is fast (~20 seconds), but website generation takes ~8 minutes for all websites. Do NOT assume the job is stuck — poll every 15-20 seconds and be patient during the website_generation step
  • Email warmup: new SMTP accounts start at 5/day, ramping to 25/day over 4 weeks
  • Voice campaigns must be: created (draft) -> activated -> triggered
  • The trigger action initiates up to 10 calls at a time, costing 1 credit each
  • Dynamic variables ({{business_name}}, etc.) are auto-populated from lead data
  • Duplicate email detection prevents re-sending to the same lead
  • Video features (with_video, with_vsl) are Pro plan only
  • Phone numbers and voice agents have a hard limit of 3 each
  • Always check preconditions before taking action (quota, credits, SMTP, plan)

平台分布

OpenClaw

87%
按下载量换算11,076

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

未展示

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。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