Browse Skills

eCTD Cross-Module Verifier

    Get Started with This Skill

    Overview

    Verify consistency and data integrity of eCTD pharmaceutical dossiers. Cross-checks Module 2 (Summaries) against Module 5 (Clinical Study Reports) to ensure study references, p-values, N-counts, endpoints, and safety data are aligned before regulatory submission to FDA, EMA, or PMDA.

    The skill targets the most common source of regulatory deficiency letters: inconsistencies between summary claims and source data. It addresses the ICH CTD structure (Modules 1–5) used globally by FDA, EMA, PMDA, Health Canada, and TGA.

    • Cross-module integrity — Every study cited in Module 2 summaries must have a matching CSR in Module 5, and vice versa.
    • Data synchronization — p-values, N-counts, and incidence rates in summary tables must exactly match the source CSR data.
    • Regulatory compliance — Checks for ICH GCP (E6), ICH E3, ICH M4, and FDA 21 CFR Part 312 references.

    Getting Started

    Add your eCTD dossier content to the project workspace — either as a single structured markdown file with Module 2 and Module 5 sections, or as a directory with module-2/ and module-5/ subdirectories. Then activate the eCTD Cross-Module Verifier skill. The agent will guide you through cross-module verification, data sync validation, and report generation.

    Example Conversation

    Below is an example of how a typical cross-module verification session unfolds. The agent verifies dossier structure, checks data consistency, and helps resolve discrepancies.

    You: I need to verify our eCTD dossier for compound RKN-4012 before we submit the NDA to FDA. The dossier files are in the workspace under ectd-rkn4012/.

    Agent: I’ll run a cross-module integrity check on your dossier. Let me start by scanning the module-2/ and module-5/ subdirectories.

    I found 3 files in module-2/ and 5 files in module-5/. Running the cross-module verifier now.

    The verification found 2 errors and 3 warnings:

    • ERROR: Study “RKN-4012-301” is referenced in Module 2.7.3 (Clinical Efficacy Summary) but has no corresponding CSR in Module 5.3.
    • ERROR: “Deaths / mortality” is mentioned in Module 2.7.4 safety summary but not found in any Module 5 CSR.
    • WARNING: Study “RKN-4012-PK-01” exists in Module 5 but is not referenced in Module 2 summaries.
    • WARNING: ICH E3 CSR structure reference not found in dossier.
    • WARNING: Endpoint “Overall Survival at 24 months” in Module 2 not matched in Module 5 endpoints.

    The most critical issue is the missing CSR for study RKN-4012-301. Do you have that report available?

    You: Yes, I forgot to add it. I’ve placed the CSR at module-5/rkn4012-301-csr.md. Can you re-run and also check the data sync between the efficacy summary and this CSR?

    Agent: Re-running cross-module verification… The missing CSR error is resolved. Now running the data sync validator between your Module 2 efficacy summary and the new CSR.

    The data sync found 1 mismatch:

    • ERROR: p-value 0.0023 (summary line 47, primary endpoint ORR) → source CSR reports p=0.0021 (line 312).

    This looks like a transcription rounding difference. The CSR uses 4 decimal places (0.0021) while the summary rounds to 0.0023. Should I update the summary to match the CSR value?

    You: Yes, use the CSR value — it’s the source of truth.

    Agent: Updated. Re-running validation — all clean now: 0 errors, 1 warning remaining (the PK study not yet summarized in Module 2). I’ve generated a verification checklist at output/ectd-verification-checklist.md for your records.

    Sample Output Excerpt

    Below is a representative excerpt from a cross-module verification report, showing the key sections reviewers care about most — study reference alignment and data consistency.


    eCTD Cross-Module Verification Report — RKN-4012

    Module 2 content: 45,230 chars | Module 5 content: 187,450 chars Studies in Module 2: 4 | Studies in Module 5: 5

    Section Completeness

    Module 2 SectionStatus
    2.4 — Nonclinical OverviewOK
    2.5 — Clinical OverviewOK
    2.7.3 — Summary of Clinical EfficacyOK
    2.7.4 — Summary of Clinical SafetyOK
    2.7.6 — Individual Study SummariesOK
    Module 5 SectionStatus
    5.2 — Tabular ListingOK
    5.3 — Clinical Study ReportsOK
    5.3.5 — Efficacy and Safety ReportsOK

    Cross-Module Consistency

    CheckResult
    Study RKN-4012-201 (Module 2 → Module 5)MATCH
    Study RKN-4012-301 (Module 2 → Module 5)MATCH
    Study RKN-4012-302 (Module 2 → Module 5)MATCH
    Study RKN-4012-PK-01 (Module 5 only)WARNING — not summarized in Module 2
    AE data: Module 2.7.4 ↔ Module 5 CSRsMATCH
    SAE counts: Module 2.7.4 ↔ Module 5 CSRsMATCH
    Deaths: Module 2.7.4 ↔ Module 5 CSRsMATCH

    Data Sync: Efficacy (Study RKN-4012-301)

    Data PointSummarySource CSRStatus
    ITT N (treatment)N=287N=287MATCH
    ITT N (placebo)N=291N=291MATCH
    ORR p-valuep=0.0021p=0.0021MATCH
    PFS HR (95% CI)0.62 (0.49–0.78)0.62 (0.49–0.78)MATCH
    OS median (treatment)18.4 months18.4 monthsMATCH

    Result: PASS — cross-module integrity appears sound.

    Built-in Validation Tools

    The skill includes two extension tools that automate the most labor-intensive parts of eCTD dossier QC.

    What the tools check

    Check Categorycross_module_verifierdata_sync_validator
    Module 2 section completenessAll sections 2.2–2.7.6
    Module 5 section completenessAll sections 5.2–5.3.5
    Study reference alignmentModule 2 ↔ Module 5 cross-refs
    Endpoint consistencyPrimary/secondary endpoint matching
    Safety data cross-referencesAEs, SAEs, deaths, discontinuations
    Regulatory citationsICH GCP, E3, M4, 21 CFR 312
    p-value matchingSummary vs. source CSR p-values
    N-count matchingSummary vs. source sample sizes
    Percentage cross-checkDetects transcription issues
    Unfilled placeholdersFlags remaining markers

    Severity levels

    • ERROR — A data mismatch or missing study/section that must be resolved before submission.
    • WARNING — A potential gap or inconsistency that should be reviewed depending on context.
    • INFO — Unfilled placeholders or source data not referenced in summaries (informational).

    Example validation output

    ======================================================================
    eCTD CROSS-MODULE VERIFICATION REPORT
    ======================================================================
    Path: ectd-rkn4012/
    Module 2 content: 45,230 chars
    Module 5 content: 187,450 chars
    Studies in Module 2: 4
    Studies in Module 5: 5
    
    --- SECTION COMPLETENESS ---
      All expected Module 2 and Module 5 sections found.
    
    --- CROSS-MODULE CONSISTENCY ---
      [WARNING] Study "RKN-4012-PK-01" present in Module 5 but not
                referenced in Module 2 summaries.
    
    --- REGULATORY REFERENCES ---
      Key regulatory references found.
    
    --- UNFILLED PLACEHOLDERS ---
      Found 3 unfilled placeholder(s).
      [INFO] line 24: Unfilled placeholder: [SPONSOR NAME]
      [INFO] line 31: Unfilled placeholder: [SUBMISSION DATE]
      [INFO] line 89: Missing value: [VALUE]
    
    --- SUMMARY ---
      Errors:   0
      Warnings: 1
      Info:     3
    
    RESULT: REVIEW NEEDED — check warnings for potential gaps.
    ======================================================================
    

    The agent runs both tools iteratively — first the cross-module verifier for structural integrity, then the data sync validator for each study pair — and surfaces findings inline so you can resolve issues before submission.

    Ready to let your expertise drive the workflow?

    Stop wrestling with rigid templates and complex tooling. Write your process in markdown, let the agent handle the rest.

    Get Started