ats-optimization

Check a resume for ATS-hostile formatting, keyword gaps, and narrative weaknesses, then propose specific inline fixes. Invoke resume-coach to perform the analysis.

This skill fires automatically when a resume has just been generated (after tailor-resume assembles a draft, before the user confirms the save) or when a resume is pasted into the conversation. It also fires on explicit request.

Workflow

1. Identify the resume

Check for a resume in context — in order of priority:

1. A resume just generated by tailor-resume in this session
2. Content pasted directly into the conversation
3. An artifact name or filename the user referenced — look it up in {user_dir}/CareerNavigator/artifacts-index.json and read the file

If no resume is identifiable:

"Which resume should I check? Paste it here or give me the filename."

2. Get the job description

A JD is required for keyword match analysis. Check for one in context:

• If tailor-resume just ran, the JD is already in context — use it
• If the user pasted a JD alongside the resume — use it
• If no JD is in context, ask:

  "Paste the job description so I can check keyword coverage. Skip it and I'll check formatting and narrative only."

If the user skips, proceed with formatting and narrative checks only — note that keyword match will be omitted from the results.

3. Invoke resume-coach

Pass the resume text and JD (if available) to resume-coach. The agent applies its ATS Scoring Rules and Narrative Coaching Rules and returns a list of issues: the check that failed, the offending text, and a proposed fix.

4. Present inline fix suggestions

Organize findings into a prioritized list — highest-impact issues first. For each issue, show the original and the proposed rewrite side by side:

ATS check — {filename or "this resume"}
{n} issue(s) found

1. {Check name}  [HIGH / MEDIUM / LOW]
   Original:  "{exact text from the resume}"
   Fix:       "{specific proposed rewrite}"

2. {Check name}  [HIGH / MEDIUM / LOW]
   Original:  "{exact text}"
   Fix:       "{specific rewrite or instruction if no direct replacement is possible}"

...

Priority levels:

• HIGH — likely to cause ATS rejection or score below 70: missing must-have keywords, non-standard section headers, multi-column layout, missing contact info
• MEDIUM — meaningfully hurts score: weak verb openers, achievements with no metrics, summary that doesn't mirror JD language
• LOW — polish items: inconsistent date formats, minor phrasing improvements

Cap the list at 10 issues. If there are more, note it:

"{n} additional low-priority items found. Address HIGH and MEDIUM items first."

If no issues are found:

"No ATS issues found. This resume is clean for parsing and keyword coverage."

5. Apply fixes

After presenting the list:

"Apply all fixes, or tell me which numbers to apply. Say 'skip' to leave as-is."

If the user confirms all or specific items:

• Apply the accepted fixes to the resume text
• Show the updated sections inline so the user can review the result
• If this resume is a saved artifact, ask:

  "Save the updated version? It will overwrite {filename}."

  • If yes: write the updated content to the file and update artifacts-index.json with today's date
  • If no: display the corrected text for the user to copy manually

If the user skips:

• Do not modify anything
• Confirm:

  "No changes made. Run /career-navigator:tailor-resume to regenerate with a fresh pass."

6. What this skill does not do

• Does not produce a score card — use /career-navigator:resume-score for a formal 40/35/25 score
• Does not fabricate experience, metrics, or skills — if a metric is missing, the fix will say "add your number" not invent one
• Does not ask the user to provide information already in context