More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Website Audit → Grounded Cold Email

    1

    Finds local businesses with dated websites, audits each one, and drafts a short cold email that quotes something real from their site.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Website Audit → Grounded Cold Email

    Website Audit → Grounded Cold Email

    Example session with this skill installed

    Find me 10 dentists in Leeds, UK with weak websites and draft cold emails I can send. Sign as Tomas.

    • Read your context and instructions
    • Compiled the website audit grounded
    • Generated the UI component

    Summary

    Searched

    Leeds, West Yorkshire, England on OpenStreetMap: 30 dentists with a website. Dropped 1 hospital, 2 chains and 1 directory link, then audited the rest one by one.

    6 drafts ready, 2 leads with no email (call list), 2 skipped because their sites are already in good shape.

    The businesses in this example are fictional. The shape of the output is exactly what a real run produces.

    leads.csv (excerpt)

    businessemailscoretop findingstatus
    Parkside Dentalparksidedentalpractice@outlook.com4homepage form has 10 fieldsdraft ready
    Oakwood Smile Studioinfo@oakwoodsmile-example.co.uk7phone number not tap-to-call, no mobile viewportdraft ready
    Kirkgate Dental Carenone found6© 2014 in the footer, jQuery 1.xno email
    Headingley Dentalhello@headingleydental-example.co.uk2site in decent shapeskipped

    emails.md: Parkside Dental

    Hook: the one high finding, the 10-field form. The medium and low findings are not stated as fact. No owner name was found, so the greeting has no name. A second address on the page belongs to another practice, so it isn't used.

    Subject: Your appointment form
    
    Hi,
    
    The appointment form on your homepage asks for 10 fields before anyone can send it. On a phone that is a lot of typing for someone who just wants a check-up, so a fair share of them give up and ring another practice in Roundhay instead.
    
    It's a sign the whole site could use a refresh, not just the form. When did you last rebuild it?
    
    Tomas
    

    check result:

    passed (grounded in "10", the field count on their own form · 66 words · one question · no dashes or exclamation marks · no banned phrases).

    Follow-up (send ~4 days later if no reply, check --followup passed):

    One more thing, the headline reads "Gentle Dentistry For All The Family Dentist in Roundhay", which looks like two lines merged. Worth a 10 minute call about the whole site?
    
    Tomas
    

    Call or contact form

    Kirkgate Dental Care, 0113 496 0123. Opener: "Your footer still says © 2014, so I had a look at the rest of the site…"

    Next steps

    1. Read each draft and send it from your own mailbox, a few per day. Your normal signature goes under the first name.
    2. Read references/compliance.md first. In the UK, sole traders count as individuals, so phone those instead.
    3. Want more? I can widen the area to Bradford or add orthodontists.

    leads.csv

    TSX · React component

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    For web designers and small agencies who want redesign clients from local businesses.

    Give your agent a niche and a city ("dentists in Leeds", "hair salons in Bristol"), or a list of URLs. It will:

    1. Find local businesses that have a website via OpenStreetMap. Free, no API key.
    2. Audit each site with a bundled script. It checks for:
      • not mobile-friendly
      • a phone number you can't tap to call
      • 7+ field contact forms
      • obsolete HTML, decade-old jQuery or Bootstrap
      • placeholder "Lorem ipsum"
      • broken HTTPS
      • a stale copyright line
      • missing reviews, and more
    3. Skip what isn't worth it: sites that are already good, dead domains, chains, and domains that have been taken over by spam.
    4. Find the contact email from the site, and the owner's first name only when it's confident. A wrong name is worse than none.
    5. Draft a 50–80 word email and a short follow-up. Each is built on one real problem and what it costs the owner in customers, and ends with a question about the whole site.
    6. Check every draft before you see it. Each one must contain something that is really on the prospect's page, and the script verifies every quoted phrase against the page. It also enforces the length, one question, no dashes and no exclamation marks, and bans phrases like "I came across" and "I hope this finds you".

    You get leads.csv (every lead with a status and the reason for any skip) and emails.md (drafts and follow-ups, plus a call list for leads with no email).

    It never sends anything. You read the drafts and send them from your own mailbox.

    Why it's different

    Most cold-email skills write from what you tell them. This one reads the prospect's site first, and won't hand you a draft it can't back up from the page.

    The rules come from a pipeline that runs live cold outreach for a web studio. It works in English and Slovak, and other languages use the same rules.

    Requirements

    • Python 3.9+ and pip install beautifulsoup4. No API keys.
    • The script reads the websites you audit and queries OpenStreetMap. Nothing else.

    Not included

    • Sending email, or sequencing.
    • Email lookup beyond what's on the business's own site.
    • Screenshots, and paid search data.

    Responsible use

    B2B only, a few emails a day, and honor opt-outs. Compliance notes are included, including the UK rule for sole traders. Not legal advice.

    • Lead finding uses OpenStreetMap, so it only finds businesses that list a website there. Coverage varies by city; you can always paste your own URLs instead.
    • The public Overpass server can be slow (up to a minute) or busy at peak times.
    • Email discovery only reads the business's own site. Sites with just a contact form end up on the call list instead of getting a draft.
    • The audit reads the HTML, not a rendered screenshot, so heavy JavaScript builders can hide some problems. Low-confidence findings are never stated as fact.
    • It drafts emails; it never sends them. Sending and compliance (GDPR/PECR, CAN-SPAM, CASL) stay with you.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 3 days ago

    Requires

    Python 3.9+
    beautifulsoup4 (pip install beautifulsoup4)
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access

    File Scopes

    ./.waa_cache/**
    ./audit_*.json
    ./draft_*.txt
    ./followup_*.txt
    ./leads.csv
    ./emails.md

    Terminal: the agent runs the bundled Python script (scripts/audit_site.py). Network: read-only GET requests to the business websites being audited (homepage, robots.txt, up to 3 contact/about pages, robots.txt respected), plus nominatim.openstreetmap.org and overpass-api.de to find local businesses. Nothing about the user or their files is sent anywhere. Files: the script writes only to ./.waa_cache/ in the working folder, and the agent writes audit JSON, drafts, leads.csv and emails.md there. No environment variables, credentials or code downloads.

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    This skill identifies local businesses with outdated or broken websites and generates personalized outreach emails. By grounding its drafts in real-time technical audits, it replaces generic templates with fact-based messages that highlight specific design and conversion flaws.

    The skill includes the primary AI workflow, a specialized `audit_site.py` script for technical diagnostics, and the necessary configurations for OpenStreetMap API integration. You will receive all the logic required to discover leads, audit their sites, and generate validated email drafts.

    You will need a local environment running Python 3.9+ with the `beautifulsoup4` library installed to handle site parsing and technical audits. It is compatible with most modern AI agents that support local script execution and external API calls via OpenStreetMap.

    No, the skill is designed to produce ready-to-send drafts and audit reports locally. This allows you to review the grounded facts and personalized hooks for quality assurance before manually sending them through your preferred email provider.

    The skill uses OpenStreetMap’s Nominatim and Overpass APIs to find businesses with active web domains in your chosen city and niche. This ensures you are targeting real, local entities rather than generic, scraped lists from outdated databases.

    Yes, it includes a validation step that cross-references the generated email text with the audit report. This ensures the drafted quotes and technical findings are 100% accurate and that the tone remains professional before the draft is finalized.

    New in Sales & Lead Gen