A CV can look finished and still be inconsistent in ways that matter the next time you edit it. Have AI audit one before changing anything, then decide which findings are formatting fixes and which need the author.
The CV
Download the workshop CV (Word, four pages). It expands the fictional Alexandra Ruiz sample CV with invented publications, positions, and institutions. It is practice material, not a record of anyone’s qualifications.
It is built to look ordinary at first glance. It contains small inconsistencies in the text and some older Word structures that you can’t see on the page.
Before you run anything, read the first page yourself and note what you would ask its author. You’ll want something to compare the tool’s findings against.
First pass: a broad audit
Attach the Word file to a fresh conversation.
Audit this CV without editing it.
Look for inconsistencies within comparable entries, missing information, possible duplicate entries, and statements that may conflict. For each finding, quote the evidence and identify the section and entry. Explain why it matters.
Separate demonstrable inconsistencies, questions for the author, and optional style preferences. Do not invent missing facts, change publication statuses, or assume that an unusual format is wrong.
Prioritize the five most useful findings, then list any others briefly. State whether you examined the document's appearance, its extracted text, or its underlying Word structure, and what you could not inspect.Attach the Word file to a fresh conversation, then send this. The last paragraph asks the tool to say whether it looked at the page, its text, or its Word structure.
Here is a recorded reply:
CV audit: Alexandra Ruiz, “alexandra-ruiz-cv-workshop.docx”
What I examined
I examined only the extracted text, as it was converted from the .docx. I did not see the rendered appearance or the underlying Word structure. I could not check:
- fonts, spacing, tab stops, and italics that the conversion may have dropped or added
- whether headings are real Word heading styles
- comments, tracked changes, hidden text, headers and footers beyond the page-1 footer line
- where the hyperlink actually points, beyond the URL shown in the text
- the blank table row after the 2020 conference entry (it may be a layout artifact)
Anything below about italics, headings, or tables is therefore a question to verify in Word, not a confirmed defect.
Top five findings
1. Same title listed as both “Forthcoming” and “in preparation” (potential conflict, question for author)
- Publications → Articles and Chapters: “Ruiz, Alexandra. ‘Small Archives, Shared Stewardship.’ Community Heritage Review. Forthcoming.”
- Publications → Research in Progress: “Small Archives, Shared Stewardship. Article in preparation on contributor review…”
These describe the same title at two different stages. “Forthcoming” usually signals acceptance, while “in preparation” does not. A reader or search committee may see this as overstating the status, or as a stale entry. I have not assumed which status is right. The author should confirm it, then either keep one entry or reconcile the two.
2. Same title appears as a conference paper and an invited workshop (possible duplicate, question)
- Conferences and Presentations, 2023: “Virtual Museums and Public Engagement in the Digital Age.” Digital Humanities Southwest Conference, Albuquerque, NM.
- Invited Workshops, 2023: “Virtual Museums and Public Engagement in the Digital Age. Digital Humanities Southwest Conference, Albuquerque, NM.”
The title, year, venue, and city are identical. It could be two separate sessions, but as written it reads like one event counted twice. If they are distinct, the entries could say how (paper vs. workshop format). It also echoes the dissertation subtitle, which is fine but worth knowing.
3. Leftover draft note and other unfinished artifacts (demonstrable)
- Digital Projects → Migration Stories Online: “[2023 draft note: replace temporary collection link before circulation.]” This is internal author text left in a circulated document.
- Header: the visible text is “alexandra-ruiz.example” but the link target is
…/archive/2022, so it may be an outdated path. That is a question about the intended destination. - Footer: “Alexandra Ruiz · Fictional workshop CV” and cover line: “September 2026” (today is October 8, 2026). Both are probably intentional for a workshop file, but both would need updating before real use.
4. Two names for what may be one project (question for author)
- Professional Experience: “2024–Present Project coordinator, Borderlands Memory Mapping Initiative… Coordinate community review of oral history maps”
- Digital Projects: “2024–present Borderlands Memory Atlas. Project coordinator. Community-reviewed maps linking oral histories…”
The role, start date, and subject are the same, but the project names differ. If they are the same effort, using one name (or saying “Atlas, a product of the Initiative”) avoids reader confusion. If they are separate, nothing is wrong.
5. Inconsistent formatting among comparable publication entries (demonstrable)
- Author name: “Ruiz, A. ‘Teaching with Unfinished Collections.’” (2025) vs. “Ruiz, Alexandra.” in all other entries.
- Page-range dash: “(2022): 45-62” uses a hyphen; the 2026 and 2025 entries use en dashes (17–34, 61–78).
- Journal title italics: Public Humanities Methods (2024) appears without italics in the extracted text, while other journal titles are italicized. Please verify in Word.
- Missing details: the 2023 chapter has no page range, and the forthcoming entry has no year. Neither is necessarily wrong, but peers in the same list include them. I have not supplied any missing facts.
Other findings
Demonstrable inconsistencies
- Conferences: the order breaks reverse-chronological sequence (2026, 2025, 2024, 2023, then 2024, 2023, 2022…). The 2021 entry uses straight quotation marks, while the others use curly ones. The first six entries are tab-separated lines and the last four are in a table, with an empty trailing row.
- Professional Experience: “2024–Present” (capital P) vs. “2021–present”; “2019-2020” uses a hyphen where its neighbors use en dashes. The section heading is bold text, whereas other sections use heading markup (verify in Word).
- Publications: “Publications” and its subsection “Articles and Chapters” appear at the same heading level in the extracted text (verify in Word).
- Service and Public Engagement: “2023–2025 Peer mentor” follows entries starting 2021–2024, so it breaks the pattern of ordering by start date. The final entry, “Member, Digital Public History Working Group,” has no dates and no bold, unlike its neighbors.
- Locations: Some entries give cities (“El Paso, TX”, “Las Cruces, NM”) while comparable ones don’t (Desert Community Archives and SSU roles, the Teaching and Learning Center workshop, the AHA meeting, THATCamp).
Questions for the author
- Is “2021–present” for the research assistantship still accurate alongside the 2026 Dissertation Completion Fellowship? This isn’t necessarily a conflict.
- No expected completion date is given for the Ph.D., if you want one.
- Is “Desert University Press” the intended imprint name, given the school is “University of the Desert”?
Optional style preferences
- Only one award lists an amount (“$4,500”). You could list amounts for all, or none.
- Digital Projects entries have no URLs, which could be added if the projects are public.
- Teaching entries for 2018–2019 repeat the Experience entry. That cross-listing is common and not an error.
A fresh incognito conversation at the default settings, with only the Word file attached. Claude converted the file to text when it was attached, and the reply says so before anything else.
Start with what it says it examined. This run read only the text Claude extracted from the file, and said so before anything else. That text still showed the table in the conference list and many of the inconsistencies, but not fonts, spacing, or Word’s heading styles, so the reply marks anything about those as something to verify in Word. Then check two or three of its findings against the document yourself.
Second pass: the Word structure
Now inspect the actual Word document structure, if your tools permit it. Do not edit the file.
Check for tables used to arrange ordinary text, empty or merged cells, headings implemented with direct formatting, inconsistent styles or local formatting overrides, hidden text, proofing-language differences, and hyperlink destinations.
Report only structures you can actually inspect. For each finding, identify its location, give the observed property, explain its practical effect, and suggest how I can verify it in Word. Distinguish a maintenance concern from a defect. If you only have extracted text, say which checks are unavailable; do not infer hidden structures from the layout.This one names what to check, so keep its findings separate from what the first pass noticed on its own. If your tool only reads the extracted text, it should say which checks it couldn't do.
This prompt names the things to look for, so it is a guided check rather than a second independent audit. Keep its findings separate from what the first pass noticed on its own.
In the recorded run, Claude said it couldn’t inspect the Word structure, listed the checks it couldn’t do, and pointed to places worth verifying in Word. It also corrected two details of its own first reply.
The recorded reply
I can’t inspect the Word structure in this conversation. The file reached me only as converted text, with no path to the original .docx, and I have no tool here that opens a .docx and reads its XML. If the file is in Google Drive, SharePoint, or Dropbox I could try fetching it, but those connectors generally return text rather than styles or XML, so I wouldn’t expect it to answer these questions. I’m not going to guess at hidden structures from the layout.
Checks I cannot do
I can’t check any of the following:
- Whether headings use Heading styles or direct formatting.
- Local formatting overrides, style names, and style consistency.
- Hidden text, comments, and tracked changes.
- Proofing language (per run, paragraph, or style).
- Whether tables are real Word tables, and whether any cells are merged or empty.
- Hyperlink targets as stored in the file, and any bookmarks or fields.
- Tab stops, indents, and spacing used for date alignment.
What the converter output suggests
None of this confirms anything about the Word file. The converter’s output points to places worth checking:
- Contact block under the title: it came out as a one-row, two-cell table. If that is a real table used for layout, it is a maintenance concern rather than a defect, since it can affect reading order for screen readers and résumé parsers. To check, click in that block and see whether Table Design and Layout tabs appear.
- Conferences and Presentations: the final entries (2022, 2021, 2020) came out as a table, followed by a one-cell empty row, while the earlier entries came out as tab-separated lines. A correction to my first reply: that is three entries in the table plus an empty row, not “the last four entries.” If it is a real table, the empty row is a small defect. It may also be a converter artifact or merged cell. To check, click into the table and look at Layout > Properties, or turn on View > Gridlines.
- Professional Experience: the heading came out as bold text, while other section titles came out as headings. This may be direct formatting standing in for a Heading style, which would make the section absent from the Navigation Pane and any generated outline. To check, click the heading and look at the style box on the Home tab, or open View > Navigation Pane and see whether the section is listed.
- Publications and Articles and Chapters: both came out at the same heading level. This may be intentional, or the subheading may have been styled at the wrong level. Check them in the Navigation Pane.
- Tab-separated date columns: dates and entries are separated by tab characters, and a few lines begin with a tab but no date (the last Service entry and the footer line). If alignment depends on tabs, the lines can drift if tab stops differ between paragraphs. To check, turn on Show/Hide ¶ (Ctrl+Shift+8) and compare the tab arrows and ruler stops across sections.
- Hyperlink: the converter shows the display text “alexandra-ruiz.example” pointing to
https://alexandra-ruiz.example/archive/2022. That is the only link it exposed. Right-click it and choose Edit Hyperlink to see the real target. In my first reply I called this a “header.” It sits in the contact block under the title, which may or may not be the page header. - “[2023 draft note: replace temporary collection link before circulation.]” is present in the converted text, so it is either visible body text or hidden or otherwise preserved text. To check, turn on Show/Hide ¶ and look for a dotted underline. Also open Review > Show Markup to see whether it is a comment or tracked insertion.
How to verify the rest in Word
- Hidden text: select all (Ctrl+A), open Font, and look at the Hidden box. A filled square means some text is hidden. File > Info > Check for Issues > Inspect Document will also list hidden content.
- Proofing language: select all and check Review > Language > Set Proofing Language. Mixed settings show nothing selected. Also check Styles pane > Modify > Format > Language.
- Direct formatting vs. styles: open the Styles pane (Alt+Ctrl+Shift+S) and Home > Styles > Reveal Formatting (Shift+F1) on a few comparable entries, such as two Experience lines or two publications.
- Comments and tracked changes: use Review > Show Markup, and check that Track Changes is off.
- Headers and footers: double-click the top and bottom margins of page 1.
If you paste what the Navigation Pane, Reveal Formatting, or the hyperlink dialog shows, I can help interpret it.
Sent next in the same conversation as the first-pass audit, which still had only the text Claude extracted when the file was attached.
A structure isn’t wrong just because it’s unusual. A borderless table, for example, is a common way to line up text. The question is whether comparable entries behave differently when you edit them.
Decide on rules
Propose a short style guide based on the prevailing patterns in this CV. Keep formatting decisions separate from factual questions that require my answer.
For each proposed change, identify the affected entries and explain what would change. Preserve names, titles, dates, institutions, publication statuses, and substantive wording. Do not remove repeated entries until I confirm they are duplicates.
Wait for me to approve the guide before editing. Any approved edit must be made in a new Word file, with a change list.Approve the guide before anything is edited. When the new file comes back, open it and look at it as well as reading the change list.
Approve the guide before anything changes. Formatting decisions are yours to make, but a conflict about a fact, such as a publication’s status, needs the author’s answer. When the new Word file comes back, open it and look at it as well as reading the change list.
The recorded guide keeps nine questions for the author apart from its formatting rules. It also says it hasn’t confirmed it can edit the Word file with its styles intact, and offers a change list you could apply yourself.
The recorded reply
Proposed style guide for the Ruiz CV
Nothing has been edited. The rules below come from patterns in the extracted text, so rules 5 and 6 depend on structure I couldn’t see and need a check in Word first.
One correction to my first audit: I said the last Service entry lacked bold “unlike its neighbors.” Only two Service entries use bold, so the inconsistency is within the whole section (rule 7).
Part A: Formatting decisions
1. Date and page ranges use an en dash with no spaces. Open-ended ranges use lowercase “present.”
- Professional Experience, “2024–Present” → “2024–present” (every other open range uses lowercase).
- Professional Experience, “2019-2020” (Public history intern) → “2019–2020”.
- Articles and Chapters, 2022 Journal of Public History and Technology article, “45-62” → “45–62”.
- Only the punctuation or capitalization changes, never the numbers.
2. Entries run in reverse chronological order by start date. This is the pattern in Education, Experience, Publications, Digital Projects, Workshops, Teaching, and Awards.
- Conferences: the 2024 “Teaching Students to Read a Digital Exhibit” and the 2023 “The Work after Launch” entries sit after the 2023 entry. Each would move up into its year group. Order within a year is your call.
- Service: “2023–2025 Peer mentor” would move to between the 2024–present and 2022–present entries.
- Only position changes, with no wording changes.
3. Your own publications use “Ruiz, Alexandra.”
- 2025, “Teaching with Unfinished Collections”: “Ruiz, A.” → “Ruiz, Alexandra.”
- Co-authored entries stay as they are.
4. Journal titles are italic.
- 2024, “Listening at Scale…”: Public Humanities Methods shows no italics in the extracted text. Please confirm in Word, and I’d italicize it only if it’s really plain.
5. Conference entries use one layout: year, tab, “Title.” Venue.
- 2022 (AHA), 2021 (THATCamp), and 2020 (Graduate Colloquium) came out as table rows, plus an empty trailing row. They would become tab-aligned lines like the first six, and the empty row would go.
- Do this only if Word confirms it is a real table and not a converter artifact. The contact block at the top is also a table, but I propose no change there because it may be intentional.
6. Section titles use the same heading style.
- Professional Experience appears as bold text and every other section as a heading. If Word confirms direct formatting, it would take the same heading style as the others.
- “Articles and Chapters” sits at the same level as “Publications.” If you intend it as a subsection, it would drop to a second level. This is your choice.
7. Bold marks the main named item in each entry. That is the role, degree, course, workshop, or project everywhere except Service. This one needs your choice:
- Option A (my recommendation): bold the organization or program name in all five Service entries. Two already do (DH Praxis Workshop, Code for Community). Three would gain bold: Digital Collections Advisory Group, Public History Graduate Student Network, and Digital Public History Working Group.
- Option B: bold the role instead. Two existing entries would lose bold on the program name.
8. Quotation marks are curly.
- 2021 THATCamp title uses straight quotes and would be converted. I’d search the file for other straight quotes in Word.
No change proposed: the “Venue, City, ST” pattern, the two-letter state abbreviations, the periods in degree abbreviations, and the 2018–2019 teaching entry that repeats the Experience entry.
Part B: Factual questions only you can answer
These stay untouched until you answer.
- “Small Archives, Shared Stewardship”: “Forthcoming” in Articles and Chapters, but “in preparation” in Research in Progress. Which status is correct? I won’t change either one.
- “Virtual Museums and Public Engagement in the Digital Age,” 2023: listed under Conferences and Invited Workshops with the same venue and city. One event or two? Both entries stay until you confirm.
- Borderlands Memory Mapping Initiative / Borderlands Memory Atlas: same project or different ones?
- Migration Stories Online: keep, delete, or resolve the “[2023 draft note…]” text, and is there a link to restore?
- Contact-block link: is the target ending in “/archive/2022” intended?
- Cover line “September 2026” and footer “Fictional workshop CV”: keep as is?
- Missing details: the year for the forthcoming article, the page range for the 2023 chapter, and a date for “Member, Digital Public History Working Group.” Locations are also absent for the Humanities Teaching Forum, Digital Stewardship Roundtable, AHA, THATCamp, and Graduate Colloquium entries. I won’t supply any of these.
- “Desert University Press”: is this the intended imprint name?
- Community Partnership Seed Grant: it’s the only award with a dollar amount. Add amounts to the others, or leave it?
Next steps
Tell me which rules to apply (all, or by number), and your choice for rule 7. The edits will go into a new Word file, with the original untouched. A change list will give each entry, its section, the before and after text, and the rule number.
I only have extracted text from your file. I haven’t confirmed that I can open the .docx and edit it in place with its styles preserved. If I can’t, I’ll say so before making anything. Then you could apply the change list in Word yourself, or I could rebuild the document from the text, which would recreate the formatting rather than preserve it.
Sent third in the same conversation. Nothing was approved or edited afterwards.
Three findings to check yours against
These were confirmed in the document when it was built. They are not output from an AI run, and the CV contains more than these three. The recorded reply above found all three, though from the text alone it could tell only that the talks sit in a table, not that the table has no borders.
A visible detail. One date range uses a hyphen, while comparable ranges use en dashes. A formatting fix: make the ranges consistent without changing the dates.
A hidden structure. Only the oldest talks sit inside a borderless table. It looks the same as the rest of the list on the page, but those entries won’t behave like their neighbors when you edit or restyle them.
An author decision. One article appears as both “Forthcoming” and “Article in preparation.” The tool shouldn’t pick one. Only the author knows which is true.
Catching every small inconsistency matters less than noticing a conflict like the last one and leaving the decision with the author.
Return to the style-guide step or the presentation.