Document Bundle Completeness Check Prompt

This prompt is designed for transaction teams, legal operations engineers, and compliance analysts who need to verify that a document bundle is complete before closing, filing, or review. The core job-to-be-done is automated first-pass completeness flagging: given a manifest or index of expected documents and the actual files provided, the prompt detects missing exhibits, schedules, amendments, and cross-referenced documents. Use this when a closing checklist, due diligence request list, or contract index exists and you need a structured gap report before human review begins. The prompt compares explicit references in the manifest against the provided file inventory and returns a list of expected-but-absent items with source citations.

This prompt requires two concrete inputs: a structured manifest listing every expected document by title, date, and reference identifier, and a complete inventory of the files actually provided. The manifest should be machine-readable—JSON, CSV, or a structured text table—not a scanned image or free-text email. The file inventory must include exact filenames, document titles extracted from first pages, and any internal reference numbers. Without both inputs in comparable formats, the prompt cannot perform reliable matching. The output is a structured gap report with three categories: present and matched, expected but absent, and ambiguous matches where fuzzy title matching produced low-confidence results. Each absent item includes the manifest reference, the expected document identifier, and a suggested resolution path.

Do not use this prompt when the manifest itself is incomplete, when documents are expected but not explicitly listed, or when the bundle contains documents referenced only inside other documents without a master index. This prompt does not interpret the substantive adequacy of documents—it only flags expected-but-absent items based on explicit references. It is not a substitute for legal judgment, and it does not verify that present documents are correctly executed, properly dated, or legally sufficient. For high-stakes transactions, always pair this automated check with human review of the gap report and manual verification of ambiguous matches. The prompt works best as a first-pass filter that reduces the manual checklist review burden, not as a final sign-off tool.

This prompt template is designed to be dropped directly into your AI harness. It instructs the model to act as a document completeness auditor, comparing a provided list of expected documents against an inventory of actual files. The model will produce a structured JSON report that flags missing items, identifies unexpected files, and provides a completeness score. The template uses square-bracket placeholders for all dynamic inputs, making it straightforward to integrate into a document processing pipeline where these values are populated programmatically from a database, a deal checklist, or a user upload.

textCopy
You are a document completeness auditor. Your task is to compare an expected document list against an actual file inventory and return a structured completeness report.

# INPUTS

## Expected Document List
[EXPECTED_DOCUMENT_LIST]

## Actual File Inventory
[ACTUAL_FILE_INVENTORY]

## Matching Rules
[MATCHING_RULES]

# OUTPUT SCHEMA
Return a single JSON object with the following structure. Do not include any text outside the JSON object.

{
  "completeness_score": <number between 0 and 1>,
  "summary": "<one-sentence summary of findings>",
  "expected_count": <integer>,
  "actual_count": <integer>,
  "matched_documents": [
    {
      "expected_name": "<string>",
      "matched_file": "<string>",
      "confidence": <number between 0 and 1>
    }
  ],
  "missing_documents": [
    {
      "expected_name": "<string>",
      "reason": "<explanation of why it was flagged as missing>"
    }
  ],
  "unexpected_files": [
    {
      "file_name": "<string>",
      "reason": "<explanation of why it was flagged as unexpected>"
    }
  ],
  "potential_matches": [
    {
      "expected_name": "<string>",
      "candidate_file": "<string>",
      "confidence": <number between 0 and 1>,
      "explanation": "<reason for low-confidence match>"
    }
  ]
}

# CONSTRAINTS
- [CONSTRAINTS]

# EXAMPLES
[EXAMPLES]

To adapt this template, replace the placeholders with concrete data. The [EXPECTED_DOCUMENT_LIST] should be a structured list, such as a markdown table or a JSON array, detailing every document required for a complete bundle. The [ACTUAL_FILE_INVENTORY] is a corresponding list of files that are present. The [MATCHING_RULES] placeholder is critical for controlling false positives; specify whether matching should be exact, fuzzy, or based on semantic similarity. Use the [CONSTRAINTS] section to enforce business rules, such as ignoring file extensions, flagging only specific document types, or requiring a minimum confidence threshold for a match. The [EXAMPLES] placeholder should contain one or two few-shot examples of a correct input-output pair to guide the model's behavior, especially for edge cases like similarly named documents or optional schedules. After adapting the prompt, always validate the output against the defined JSON schema before allowing it to proceed downstream.

The Document Bundle Completeness Check prompt is designed to operate as a critical gate in a transaction or compliance document pipeline. It should be invoked after document ingestion and classification but before downstream extraction, comparison, or summarization workflows. The prompt expects a structured input containing a manifest of expected documents (such as a closing checklist, regulatory filing index, or contract hierarchy) alongside the actual document set inventory with metadata. The output is a structured completeness report with flags for missing exhibits, schedules, amendments, and cross-referenced documents that are expected but absent from the provided bundle.

To wire this prompt into an application, build a pre-processing step that assembles the [EXPECTED_DOCUMENT_MANIFEST] and [ACTUAL_DOCUMENT_INVENTORY] from your document management system or data room index. The manifest should include document titles, expected types (exhibit, schedule, amendment, main agreement, side letter), and any cross-reference identifiers. The inventory should list actual filenames, extracted titles, document types, and page counts. Pass both as structured JSON or formatted text blocks into the prompt. After receiving the model response, run a schema validator against the expected output structure: a JSON object with completeness_status, missing_documents array (each with expected_document, reference_source, absence_evidence, risk_level), present_documents array, and unexpected_documents array. Reject malformed responses and retry with the validation error message appended to the prompt context.

For high-stakes transaction reviews, implement a human review queue for all risk_level: high flags and any missing_documents entries where absence_evidence is inferred rather than confirmed_absent. Log every completeness check run with the input manifest hash, output report, validator results, retry count, and reviewer decision. This audit trail is essential for regulatory filings, M&A closings, and compliance submissions where document completeness claims must be defensible. Choose a model with strong instruction-following and structured output capabilities (GPT-4o, Claude 3.5 Sonnet, or equivalent). Avoid using smaller or older models that may hallucinate document references or fail to maintain the output schema under long context. If your document set exceeds the model's context window, pre-filter the manifest to relevant sections and run multiple targeted checks rather than attempting a single oversized completeness run. Never treat the model's output as a final completeness certification without human verification of flagged gaps.