Run Contract Due Diligence Across a Set of Agreements

Turn a stack of target-company agreements into a quoted, per-contract issue table, a ranked red-flag list, a consent tracker and a first-draft disclosure schedule.

The Workflow

1. 1

   Set the deal context and load the first batch

   Tell Claude whose side you are on, the deal structure and the materiality threshold before it reads anything. Then paste agreements in batches, each in its own labeled document tags with a contract ID you will use everywhere else. Smaller batches (five to ten agreements) make the extraction easier to check.

   PromptCopy

   I represent [BUYER / SELLER / INVESTOR] in a proposed [STOCK PURCHASE / ASSET PURCHASE / MERGER / FINANCING] involving [TARGET COMPANY]. Signing is targeted for [DATE] and closing for [DATE]. Our materiality threshold for this review is [e.g., contracts with annual value above $X, or any contract with exclusivity or change-of-control terms].
   
   Below are [NUMBER] agreements, each in its own document tags with a contract ID.
   
   <document id="C-01" title="[COUNTERPARTY] [AGREEMENT TYPE]">
   [PASTE AGREEMENT TEXT]
   </document>
   
   <document id="C-02" title="[COUNTERPARTY] [AGREEMENT TYPE]">
   [PASTE AGREEMENT TEXT]
   </document>
   
   Before any analysis, list each contract ID with the agreement title, the parties as stated in the preamble, the effective date, and whether any exhibits, schedules or amendments it references appear to be missing from what I pasted. Flag any pages that look truncated or illegible. Do not extract terms yet.

   What you get: An inventory by contract ID with parties, dates and a list of missing exhibits or amendments. Request the missing pieces from the data room before extraction, since an amendment can change the answer on assignment or term.

2. 2

   Build the per-contract issue table

   Have Claude extract the same fields from every agreement, quoting the clause and section number for each. Fields Claude cannot find are marked 'not found' rather than inferred. This table is the core work product that everything else is built from.

   PromptCopy

   For each contract ID, build an issue table with one row per field below. Columns: Field | Finding | Quoted language | Section.
   
   Fields:
   1. Parties and any affiliates bound
   2. Term, renewal and notice-of-nonrenewal period
   3. Assignment (including any restriction on assignment by operation of law or by merger)
   4. Change of control (definition, what it triggers: consent, notice, termination right, payment)
   5. Exclusivity, non-compete and non-solicit
   6. Termination rights (for convenience and for cause), with notice periods
   7. Most-favored-nation or pricing-parity terms
   8. Limitation of liability and caps, including carve-outs
   9. Consent or notice requirements not covered above
   10. Governing law and dispute resolution
   
   Rules: quote the operative language exactly and give the section number. If a field is not addressed in the agreement, write 'not found' and do not infer a default rule. If language is ambiguous, say so in the Finding column and explain the competing readings in one sentence. Do not apply the deal structure yet.

   What you get: One table per contract with a quote and section on every populated row. Spot-check the quotes against the source; a paraphrase in the quote column is a sign to re-run that contract.

3. 3

   Rank the red flags against the deal structure

   Now apply the transaction. Ask Claude which contracts appear to require consent, give the counterparty a termination right, or trigger a payment because of how this deal is structured, and rank them. Every item carries its quoted clause and an honest statement of uncertainty.

   PromptCopy

   Using the issue tables above and the deal structure I gave you ([STRUCTURE]), produce a ranked red-flag summary.
   
   For each contract that may be affected, give: Rank | Contract ID | Counterparty | Trigger (consent, notice, termination right, payment, exclusivity conflict, other) | Why the structure may trigger it | Quoted clause and section | Confidence (clear / arguable / unclear).
   
   Rank by likely impact on the deal: first, contracts that could terminate or block closing; then those requiring consent; then notice-only items; then post-closing restrictions such as exclusivity or non-competes that would bind the combined business.
   
   Where the answer depends on how [JURISDICTION] law treats the transaction (for example, whether a merger or stock sale counts as an assignment), mark it 'unclear: depends on governing law' and do not resolve it. List any contracts above our materiality threshold that you reviewed and found no trigger in, so I can confirm the negative.

   What you get: A ranked list with the clause quoted for each item and a confidence label. The 'unclear' items are the ones to research under the governing law before advising the client.

4. 4

   Build the consent and notice tracker

   Turn the red flags into an action list the deal team can work from, with draft request language for each counterparty. Timing pulls from the notice periods in the contracts and the target dates you supplied.

   PromptCopy

   Build a consent and notice tracker from the red-flag summary. Columns: Contract ID | Counterparty | Required action (consent / notice / waiver) | Contractual timing requirement (quoted, with section) | Latest send date relative to [SIGNING OR CLOSING DATE], marked [VERIFY date math] | Form required (written, specific address, specific person) | Status.
   
   Then, for each consent item, draft a short request letter paragraph that identifies the agreement by title and date, describes the transaction at the level of detail in [APPROVED DEAL DESCRIPTION], cites the section requiring consent, and asks for written consent by [DATE]. Do not describe deal terms beyond the approved description, and keep confidentiality obligations in mind. Flag any agreement where the notice address or method is missing.

   What you get: A tracker sorted by send date with draft request language. Confirm every date calculation and the notice mechanics (address, method, recipient) against the agreement before anything goes out.

5. 5

   Draft the material-contracts and required-consents schedules

   Ask Claude for a first-pass disclosure schedule keyed to the representation in the purchase agreement. Paste the relevant representation so the schedule follows its actual categories rather than a generic list.

   PromptCopy

   Here is the material-contracts representation and the consents representation from the purchase agreement:
   
   <document title="Purchase agreement reps">
   [PASTE THE MATERIAL-CONTRACTS AND CONSENTS REPRESENTATIONS, WITH SECTION NUMBERS]
   </document>
   
   Draft a first pass of the corresponding disclosure schedules. Follow the categories in the representation exactly, list each responsive contract under the category it fits, identified by title, parties and date, and cross-reference the contract ID. For the consents schedule, list each agreement requiring consent or notice and the section that requires it.
   
   After the draft, list: (a) contracts that may fit a category but where you are uncertain, with the reason; (b) categories in the representation for which nothing in the reviewed set appears to be responsive; and (c) any category that depends on information not in these documents, such as annual spend. Mark the whole draft 'DRAFT: subject to attorney review'.

   What you get: A schedule draft organized to match the rep, plus a list of uncertain fits and empty categories. Empty categories often signal contracts that have not yet been produced.

6. 6

   Attorney review before anything leaves the deal team

   Verify every extracted term against the source contract, confirm the governing-law questions flagged as unclear for [JURISDICTION], and check the consent dates and notice mechanics yourself. Keep the issue tables and red-flag summary within the deal team, and treat the schedules as a draft until the responsible attorney signs off.

   What you get: Issue tables you have spot-checked, governing-law questions resolved by research rather than by Claude's reading, and schedules reconciled to the final purchase agreement. The output speeds the review; it does not replace it.

Example Output

Illustrative example — names, facts, and figures are fictional.

RED-FLAG SUMMARY (ILLUSTRATIVE / FICTIONAL)
Matter: Acquisition of Larkspur Analytics, Inc. by Northgate Holdings (stock purchase). We represent the buyer.

Rank | Contract ID | Counterparty | Trigger | Quoted clause and section | Confidence
1 | C-07 | Kestrel Health Systems | Termination right on change of control | "Customer may terminate this Agreement upon thirty (30) days' written notice following any Change of Control of Vendor." (Sec. 14.3) | Clear
2 | C-02 | Bayfront Cloud Services | Consent required | "Any change in the beneficial ownership of more than fifty percent (50%) of Customer's voting securities shall be deemed an assignment requiring Provider's prior written consent." (Sec. 11.2) | Clear
3 | C-11 | Orion Data Partners | Possible assignment by operation of law | "Neither party may assign this Agreement, by operation of law or otherwise, without consent." (Sec. 9.1) | Unclear: depends on governing law ([JURISDICTION]) and final structure
4 | C-05 | Tidewater Retail Group | Exclusivity binding affiliates post-closing | "Vendor and its Affiliates shall not provide the Services to any Competitor..." (Sec. 6.1) | Arguable: 'Affiliates' definition (Sec. 1.1) may sweep in buyer's portfolio

REVIEWED ABOVE THRESHOLD, NO TRIGGER FOUND
C-01, C-03, C-09 (assignment and change of control: 'not found')

CONSENT TRACKER (excerpt)
C-02 | Bayfront Cloud Services | Consent | "prior written consent" (Sec. 11.2) | Send by [DATE] [VERIFY date math] | Written, to General Counsel at notice address (Sec. 18) | Not started

DRAFT: subject to attorney review. All quotes to be verified against the executed agreements.

Tips

• •Give Claude the deal structure before the contracts. The same anti-assignment clause matters very differently in an asset purchase than in a stock purchase, and Claude cannot rank red flags without knowing which you are doing.
• •Insist on 'not found' over inference. A blank assignment field is information (the contract may be silent and default law applies), and you want to see that gap rather than have Claude fill it with a guess.
• •Work in batches of five to ten agreements and keep each batch's issue tables in the same conversation as the roll-up. Contract IDs keep the references stable across steps.
• •Check definitions, not just operative clauses. 'Change of Control', 'Affiliate' and 'Competitor' are often defined in Section 1 and change the answer; ask Claude to quote the definition alongside any trigger it flags.
• •Amendments and side letters frequently change assignment and term. If the inventory step shows a referenced amendment is missing, get it before relying on that contract's row.

Frequently Asked Questions

Prompts used in this workflow

Related Reading

• AI Contract Review: How to Use Claude for Due Diligence
• Harvey AI Contract Review: How It Works vs Claude
• Triage an Incoming Contract Against Your Standard Positions
• Corporate Law Practice