Skip to content

msdanyg/ats-resume-skill

ats-resume-formatting

Verified

Use this skill whenever a resume needs to be formatted, reviewed, or debugged for upload to Workday, Greenhouse, Lever, iCIMS, or any other ATS (applicant tracking system). Trigger on requests like "format my resume for ATS," "is my resume ATS-friendly," "why is Workday leaving fields blank," "clean up my resume for online applications," or "make my resume parse-safe." Also trigger proactively any time a user is building or editing a resume that will be uploaded to a job portal, even if they don't mention "ATS" by name — most resume-formatting requests have this failure mode lurking under the surface. Prevents the most common cause of silent application failure — a resume that looks perfect to a human but leaves the Company, Dates, or Title field blank or wrong in the parsed record, which means the candidate never surfaces in recruiter search regardless of how strong the resume is.

FileKindSize
SKILL.mdskill md14.6 KB
references/failure-patterns.mdmarkdown5.1 KB

references/failure-patterns.md

View on GitHub
# ATS Parse Failure Patterns — Worked Examples

This reference walks through exactly how each failure mode in `SKILL.md` happens, with before/after examples. All company names below are fictional, chosen only to illustrate a naming *pattern* (lowercase brand, acquired subsidiary, domain-styled name, etc.) — not real companies.

---

## Pattern 1 — The lowercase brand

**Before (fails):**
```
flowly
Product Marketing Manager
2021 – 2024
```

**What happens:** capitalization-weighted NER treats a lowercase-initial token as a common noun, not a proper noun — `flowly` reads like a stray word, not a company. Combined with the year-only date (Pattern 3 below), this block can extract with Company blank and dates missing entirely.

**After (parses):**
```
Product Marketing Manager
Flowly
Mar 2021 – Jun 2024
```

---

## Pattern 2 — The parenthetical title line

**Before (fails):**
```
Senior Director of Marketing (Growth & Lifecycle)
Northwind Analytics
Jan 2022 – Present
```

**What happens:** the parenthetical reproduces the exact `Name (Org)` shape parsers use to detect a company name elsewhere in a block. If `Northwind Analytics` is *also* an out-of-dictionary or ambiguous token, the parser can fall back to scanning the block and seize `(Growth & Lifecycle)` as the Company field instead of the real employer two lines down.

**After (parses):**
```
Senior Director of Marketing
Northwind Analytics
Jan 2022 – Present
```
The qualifier moves to the name-block headline instead, where there's no Company field for it to contaminate:
```
Jordan Rivera
Senior Director of Marketing (Growth & Lifecycle)
Chicago, IL · jordan@email.com · linkedin.com/in/jordanrivera
```

---

## Pattern 3 — The year-only date range

**Before (fails):**
```
Marketing Manager
Northwind Analytics
2019 – 2022
```

**What happens:** this is the single most common failure across every parser tested for this skill. Some engines simply fail to map a year-only range at all, leaving the entire tenure blank in the parsed record — not just the date, the whole entry.

**After (parses):**
```
Marketing Manager
Northwind Analytics
Aug 2019 – Nov 2022
```

If you genuinely don't remember the exact month, pick a placeholder and apply it consistently (many people default to January or July) rather than leaving any range as year-only.

---

## Pattern 4 — The company/location same-line pairing

**Before (fails):**
```
Product Marketing Lead
Cascade Software, Remote
Jun 2020 – Feb 2022
```

**What happens:** the comma-separated pairing can be read as a location line rather than a company line, leaving Company empty even though `Cascade Software` is a perfectly parseable, well-known-shaped name.

**After (parses):**
```
Product Marketing Lead
Cascade Software
Jun 2020 – Feb 2022
```
Location moves to the contact block at the top of the resume, or gets entered directly in the ATS's location form field — never appended to the company line.

---

## Pattern 5 — The domain-styled or acquired brand

**Before (fails):**
```
Marketing Manager
Loopwave.io
Jan 2020 – Present
```

**What happens:** the `.io` domain styling reads to the parser as a URL fragment, not a company name — an out-of-dictionary NER miss regardless of how clean the surrounding layout is.

**After (parses), if there's a recognized parent company:**
```
Marketing Manager
Loopwave (Salesforce)
Jan 2020 – Present
```
The parenthetical here is a *recognized company name*, which is different from Pattern 2's parenthetical (a description). That distinction is the whole test: would a human recognize the word in parentheses as an actual company? If yes, it's a legitimate anchor. If it's a description (`(a Salesforce acquisition)`, `(now part of Slack)`), it belongs in a bullet instead, not the header.

**After (parses), if there's no recognized parent:**
```
Marketing Manager
Loopwave
Jan 2020 – Present
```
De-stylize to plain word form and rely on isolation (its own line, nothing after it) rather than inventing a designator that isn't part of the real legal name.

---

## Pattern 6 — The bulleted, fused education entry

**Before (fails):**
```
Education
• MSc, Environmental Science — University of Toronto
```

**What happens:** three failures stack in one line. The bullet glyph marks the line as description text rather than an entity header. The bare `MSc` acronym misses the degree-level dictionary (which maps `Master of Science`, not the abbreviation). The em dash fuses degree and institution into a single string the parser can't split. The result: this section frequently parses as nothing at all — no degree, no institution.

**After (parses):**
```
Education

Master of Science (MSc), Environmental Science
University of Toronto
```
No bullet, degree spelled out and paired with its acronym, institution on its own line beneath it.

---

## The meta-lesson

None of these failures are visible by reading the resume normally — every "before" example above looks completely reasonable to a human. That's exactly why the plain-text paste test in `SKILL.md` matters: it's the only cheap way to see the document the way the parser sees it, before you find out the hard way that a required field came through blank.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall