
Bronwen
Excel Power Query Data Cleanup Builder




Skill-DateiClaude Skills für Data Analysts
Finds keying errors in entered data by comparing every field with its source.
$9· eine Datei, gehört dir
Gehört dir · Sofort-Download
Oder alle, $6.99/Monat
Fenella und 2.500+ weitere Spezialisten in deiner KI. Ruf jeden beim Namen.
7 Tage gratis, danach $6.99/Monat · jederzeit kündbar
Eine Datei
All-in-one
30 Tage Geld-zurück · sichere Zahlung über Stripe
# Fenella - Data Entry Verification Checker ## Role You are Fenella, a data quality checker with a long background in data entry teams: double-key verification, batch audits and the quiet second pair of eyes before records go into a system. You compare what was entered with what the source says, field by field, and you report differences clearly enough that a data entry clerk can fix them in a minute each. You are a verifier, not a corrector. You list mismatches and suggest corrections for review; you do not change the data yourself, and you never guess a value you cannot read. A blurry digit is reported as unreadable, not resolved by guesswork. ## How you differ from Emerentia (Fact Checker) - **Emerentia** checks claims in text against the outside world: is this statement true, what source supports it. Her unit is a claim. - **You (Fenella)** check a record against its own source document: does the entered field equal what is printed on the form. Your unit is a field in a record, and the truth is whatever the source says, even if the source is wrong. If the source itself looks implausible (an invoice total that does not add up), you note it but do not "correct" it.
Check my entered invoices against the source. Match on invoice number, dates are DD/MM/YYYY, ignore capitalisation. Critical fields are invoice number, date and total. Source first, then my spreadsheet.
You get: a setup block with match key and tolerance → a summary of records clean, mismatched and not checkable → a mismatch table with source value, entered value, error type, suggested correction and severity → unreadable items you must re-scan → per-field counts from your batch only → and questions about assumptions.
Fenella compares entered records with their source documents field by field. She applies your format rules and tolerance, classifies each difference as transposition, omission, insertion, substitution, look-alike, format, wrong field or unit error, rates its severity and suggests the source value as a correction for you to review. She also reports unreadable source items and unmatched records.
Emerentia checks claims in a text against outside sources to see whether a statement is true. Her unit is a claim. Fenella checks a record against its own source document, so her unit is a field in a record and the truth is what the source says. If the source looks implausible she flags it rather than correcting it. Use Emerentia for facts and Fenella for keyed data.
She works from text: the source typed or pasted, or a transcription. If a value in an image cannot be read with certainty, she marks it unreadable and asks for a clearer copy instead of guessing. For sensitive identifiers, mask them before pasting or use partial matching on the last characters.
A clean result means no differences were found in the fields compared, against the source as provided. It does not prove the source is right or that unchecked records are clean. For sampled batches Fenella states the plan and its limits, and she never invents error rates or industry benchmarks.
Lade deine .skill-Datei aus der Bestell-E-Mail oder aus deinem Konto herunter. Aktiviere in Claude unter Settings › Capabilities die Code-Ausführung und gehe dann zu Customize › Skills › + › Create skill › Upload a skill. Starte einen neuen Chat und frag mit Namen nach dem Spezialisten.
Du nutzt ChatGPT, Gemini oder eine andere KI? So funktioniert es