Yutsukioka Un Job Application Helper Apex Cross Doc Consistency

apex-cross-doc-consistency

Purpose

This skill compares generated application documents against each other and against the source inputs to detect inconsistencies that could undermine the candidate's credibility during screening. It produces a structured consistency report with actionable flags.

Apply the expert lens, collaboration rules, guardrails, quality loop protocol, and guiding principles defined in apex-guardrails.

It does not rewrite documents; it flags issues with specific locations and suggested fixes.

Inputs

Required:

• At least two generated documents from Phase 8 (any combination of Option 1–6).

Optional:

• USER_JOB_HISTORY_TEXT: to verify that documents match the source data.
• TERM_EXTRACTOR: to check keyword usage consistency.
• inputs/application_context.md: for full cross-reference.

Checks

1. Job titles: Verify that the same role is titled identically across all documents. Flag any discrepancies (e.g., "Programme Manager" in the CV vs. "Project Manager" in the Cover Letter).
2. Dates and timelines: Verify that employment dates, project durations, and chronological order are consistent across documents.
3. Achievements and metrics: Verify that quantified claims (e.g., "USD 5M budget", "12 countries") match across documents and against source inputs. Flag any conflicting numbers.
4. Keyword usage: Verify that high-priority keywords (★★★ and above) appear consistently across documents. Flag any document that omits a critical keyword present in others.
5. Narrative coherence: Verify that the UVP, STAR stories, and key selling points referenced in the strategy report are reflected in the generated documents. Flag missing narrative threads.
6. Employer and organization names: Verify spelling and naming consistency across documents.
7. Format compliance (only if strict-field outputs included): If Option 1 or Option 4 outputs are included, flag bullets/line breaks/curly quotes that violate strict paste rules.

Output format

Return a structured report:

## Cross-Document Consistency Report

### PASS items
- <list of checks that passed>

### FLAGS
- [FLAG-001] <Category>: <description of mismatch, with document names and locations>
- [FLAG-002] ...

### Summary
PASS / FAIL (with count of flags)

Rules

• Do not edit or rewrite the documents; only report findings.
• Reference specific locations in each document when flagging issues.
• Distinguish between critical flags (factual mismatches) and minor flags (stylistic differences).
• If no inconsistencies are found, return a PASS report confirming the documents are consistent.

Steps

1. Parse all provided documents and extract job titles, dates, metrics, keywords, and key claims.
2. Cross-reference each extracted element across all documents.
3. Cross-reference against source inputs if provided.
4. Compile flags for any mismatches.
5. Output the structured consistency report.

Converted and distributed by TomeVault — claim your Tome and manage your conversions.