Free Claude Skill
The Lead Waterfall Skill
This is what our team uses to build outbound lists. You tell it who your dream customer is. It finds those people, checks every email, and throws out the junk. What comes back is a clean file you can upload and send. Drop it into your AI agent and let it run.
01What's inside
Two Tools, One Job
The Playbook
lead-waterfall-processing
The instruction book your AI agent follows. It sets the order for finding people (Blitz first, then Prospeo, then Ark), how to check every email, how to spot the people who don't fit your customer, and how to pack the final files. Your agent does the work. The playbook keeps it honest.
The Cleaner
sequencer-list-scrubber
A small program that cleans your list. It removes fake emails, broken emails, and addresses like info@ or sales@ that don't belong to a real person. Then it shows its work: it counts every row, checks its own output again, and writes the numbers down. When in doubt, the row goes out.
02The pipeline
Dream Customer In, Clean List Out
Ask First
The agent asks before it does anything. Who's your dream customer? How many people do you need? What does it need to know about each one? It won't guess at the answers.
Pick the Right Tool
Every search starts with Blitz. If Blitz can't find someone's email, the agent tries Prospeo, then Ark. You never pay for a lookup you don't need.
Fill in the Blanks
Every person gets a tracking number, so only the ones still missing info move on to the next tool. Nothing gets lost, and you see the counts at every step.
Check Every Email
Every email gets tested with No2Bounce. Want a second opinion? EmailListVerify runs too. Only the ones that pass move forward. Risky addresses go in their own file, so they never sneak into your send list.
Keep Only Good Fits
Now the AI reads each contact and asks whether this person is actually your customer. Right job, right industry, right place. Anyone who doesn't fit gets removed, and the reason gets written down.
Clean Out the Junk
This step runs on code, not AI guessing. It throws out fakes, typos, throwaway accounts, and addresses like info@ or billing@ that don't belong to a real person. One failed check and the row is out.
Pack and Double-Check
Clean contacts get split into upload-ready files. The agent counts everything one last time and shows you the numbers. Nothing gets uploaded until you say go.
03Why it works
Design Principles
Waterfall, not shotgun
Each tool only sees the people the last tool couldn't finish. That keeps the cost down and the list clean. That's what makes it a waterfall.
Code cleans, AI judges
Code handles email cleanup, so you get the same answer every time. The AI does one job: deciding whether a person fits your dream customer. Judgment calls need AI. Cleanup doesn't.
Proof, not promises
Every run writes a report with all the counts, plus a fingerprint for every file. The cleaner re-checks its own work before anything gets called done, so you can audit the run instead of trusting it.
04Setup
Install In Two Minutes
- 1Download the zip and unzip it. You'll see two folders:
lead-waterfall-processingandsequencer-list-scrubber. - 2Put both folders in your agent's skills folder. For Claude Code, that's
~/.claude/skills/. Other AI agents work the same way. - 3Give your agent a lead list, or describe your dream customer, and ask it to run the lead waterfall. The cleaning step only needs Python 3, no accounts and no keys. To find new people, the agent uses your accounts with the finder tools.
The skill names the tools we use. If you use different ones, swap them in. The process works with any stack.