agentleFS
Sign inStart free
prosyo-aiprosyo-ai/gtm-skill/skills/list-cleaning/SKILL.md

list-cleaning

Skill13 starsChanged 22 days agoMITView on GitHub

Use this skill when a prospect or contact list needs cleaning before outreach or import: \"clean this CSV\", \"dedupe my leads\", \"fix the names and titles\", \"why are my bounces high\". Normalises names, titles, companies and domains, removes duplicates, invalid and risky emails, role addresses, customers and opt-outs, flags stale contacts, and outputs a send-ready list with a cleaning report. Protects sender reputation and stops embarrassing merge-field mistakes.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this before any list touches a sequence or a CRM import.

    line 16
  2. Distinctive:

    Titles: keep the original, add a normalised seniority and function column.

    line 32
  3. Distinctive:

    Read references/cleaning-rules.md for the full normalisation and risk rules with edge cases.

    line 35
  4. Distinctive:

    Remove customers, open deals, opt-outs and anyone contacted in the last 90 days.

    line 44
  5. Distinctive:

    MUST keep the original values alongside normalised ones.

    line 68
  6. Distinctive:

    MUST output a removed-rows file with reasons.

    line 69
  7. Distinctive:

    NEVER delete records from the user's CRM.

    line 70
  8. Distinctive:

    NEVER guess an email address and mark it as verified.

    line 71

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Use this before any list touches a sequence or a CRM import. It produces a clean list and a report of exactly what was removed or changed and why.

Inputs

  • The raw list.
  • Suppression sources: customers, open opportunities, opt-outs, do-not-contact, recent contacts.
  • The campaign it is for, so you know which columns must be filled.

The play

1. Profile the file

Count rows, columns, blanks per column, and obvious junk (test rows, header repeats, rows with only an email). Note the encoding problems: broken accents in names are common and visible in a first line.

2. Normalise

  • Names: proper case, strip titles and suffixes (Dr., PhD, MBA), split full names, fix all-caps. Keep particles like "van", "de", "Mc" correct.
  • Companies: strip legal suffixes for display (Inc., Ltd, GmbH, LLC) while keeping the original in another column. "Acme Holdings Inc." becomes "Acme" in an email.
  • Titles: keep the original, add a normalised seniority and function column.
  • Domains: lower case, strip www and paths, derive from email when missing.

Read references/cleaning-rules.md for the full normalisation and risk rules with edge cases.

3. Deduplicate

Match on email first, then on LinkedIn URL, then on first name + last name + domain. Keep the most complete record and merge the rest.

4. Remove risk

Drop invalid emails, disposable domains, role addresses (info@, sales@, admin@, support@), personal domains for B2B campaigns, and anything that bounced before. Flag catch-all domains rather than dropping them; decide by campaign size.

5. Suppress

Remove customers, open deals, opt-outs and anyone contacted in the last 90 days. Match suppression on email and on domain where the list is account-level.

6. Report

Show what happened to every removed row and hand back a list of rows that need a human decision.

Output format

Input: N rows  →  Output: N send-ready rows
Fixed: N names, N company names, N domains
Removed: N duplicates | N invalid | N role | N personal | N disposable | N suppressed (by reason)
Flagged for review: N catch-all | N missing first name | N possible job change
Files: clean list, removed rows with reason, review list

What good looks like

  • A random 20-row sample reads perfectly in a merge preview: "Hi Ana," at "Northwind", never "Hi ANA" at "Northwind Holdings, Inc.".
  • Removed rows keep their reason, so nothing disappears silently.
  • Bounce rate on the cleaned list stays under 2%.
  • The common mistake is deduplicating only on email, which leaves the same person in the list twice with a work and an old address.

Rules

  • MUST keep the original values alongside normalised ones.
  • MUST output a removed-rows file with reasons.
  • NEVER delete records from the user's CRM; cleaning produces new files or proposed changes for approval.
  • NEVER guess an email address and mark it as verified.

Example prompts

  • "Clean this 3,500-row CSV for a cold email campaign."
  • "Dedupe these two lists and remove our customers."
  • "Fix the names and company names so merge fields look natural."

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free