Contract Playbook Builder
When your team keeps renegotiating the same clauses from scratch and you want a written standard, with fallbacks and escalation rules, that Claude and non-lawyers can triage against.
Most legal teams already have a playbook. It just lives in the heads of two senior lawyers and in a folder of signed deals. Writing it down pays off quickly: junior lawyers and business teams can triage incoming contracts consistently, escalations become predictable, and you stop relitigating the same liability cap with every vendor. The hard part is starting, because a useful playbook has to reflect what you actually sign, not an idealized wish list.
This prompt builds the first draft from evidence. You paste a set of signed agreements of one type and state your preferences, risk tolerance and approval levels. Claude organizes the result by clause category, giving model language for the standard position, ranked fallbacks, unacceptable terms and an escalation trigger, and it quotes the signed agreements that support each position. Where your past deals disagree, it shows the range and asks which to adopt, and it labels positions that come only from your stated preferences.
A playbook encodes judgment, and judgment belongs to the attorneys who approve it. Claude can only infer positions from the agreements you supply, and past deals may reflect one-off concessions or outdated law. Review every model clause, decide the open questions, and confirm escalation rules with whoever owns them. Once approved, the playbook can sit in a Claude Project as reference material, but outputs built on it still need attorney review.
The Prompt
<signed_agreements> <document name="[AGREEMENT 1, e.g., Vendor A MSA 2024]"> [PASTE TEXT] </document> <document name="[AGREEMENT 2]"> [PASTE TEXT] </document> [ADD 3-10 SIGNED AGREEMENTS OF THE SAME TYPE] </signed_agreements> <our_preferences> Organization: [COMPANY OR FIRM / CLIENT NAME] Agreement type this playbook covers: [e.g., vendor SaaS agreements where we are the customer] Our usual side: [CUSTOMER / VENDOR / LICENSOR / etc.] Risk tolerance and priorities: [e.g., data security is non-negotiable; we flex on payment terms] Approval levels: [WHO CAN APPROVE WHAT, e.g., legal ops can approve fallbacks; GC approves anything off-playbook] Positions I already know I want: [LIST ANY] </our_preferences> Build a contract playbook for this agreement type, based on what we actually agreed to in the signed agreements and on the preferences above. For each clause category (limitation of liability, indemnity, IP, confidentiality, data protection, warranties, term and termination, payment, governing law and disputes, assignment, insurance, and any other category that appears in the agreements): 1. Standard position: our preferred language, drafted as model clause text. 2. Acceptable fallbacks: up to two, in order of preference. 3. Unacceptable terms: what we reject. 4. Escalation trigger: what sends the contract to [APPROVER]. 5. Evidence: quote where each position appears in the signed agreements, by document name and section. Where the agreements disagree, show the range and ask me which to adopt. Mark any position that comes only from my stated preferences (not from the signed agreements) as "Preference only." List clause categories where the agreements give no guidance. Format the result as a single document with headings by clause so it can be saved into a Claude Project as reference material. This playbook is a draft for attorney review before anyone relies on it.
Example Output
A clause-by-clause playbook with model language, ranked fallbacks, rejected terms, escalation triggers and quotes from the signed agreements that support each position.
Illustrative example — names, figures, and facts are fictional.
CONTRACT PLAYBOOK: Vendor SaaS Agreements (Company as Customer) Organization: Meridian Outfitters, Inc. (fictional) | Approver for off-playbook terms: General Counsel Source agreements: 6 signed vendor MSAs, 2022-2025 1. LIMITATION OF LIABILITY Standard position: "Each party's aggregate liability shall not exceed the fees paid or payable in the twelve (12) months preceding the claim. This limit does not apply to breaches of Confidentiality, Data Security, or a party's indemnification obligations." Fallbacks: (1) Same cap, with data-security breaches subject to a super-cap of 2x annual fees. (2) Cap of fees paid in the prior 12 months, no carve-outs, only for vendors processing no personal data. Unacceptable: Any cap below 6 months of fees; caps that apply only to the vendor. Escalate to GC: Any data-security carve-out removed for a vendor processing customer personal data. Evidence: - Fennick Analytics MSA, Sec. 10.1: "...fees paid in the twelve (12) months preceding..." (standard) - Calloway Cloud MSA, Sec. 9.2: super-cap "two times (2x) the annual fees" (fallback 1) - RANGE NOTE: Tessaro HR MSA, Sec. 11 accepted a 6-month cap. Was this a one-off? Please confirm whether 6 months should be acceptable or escalate-only. 2. DATA PROTECTION Standard position: Vendor signs our data processing addendum; breach notice within [NUMBER] hours of discovery. Fallbacks: (1) Vendor's DPA if it includes audit rights and subprocessor notice. Unacceptable: No breach-notice obligation. Escalate: Any transfer of personal data outside [APPROVED REGIONS]. Evidence: Breach-notice periods in the signed agreements range from 48 to 72 hours (Fennick Sec. 7.3; Calloway Sec. 6.4). Preference only: Audit rights (none of the six agreements include them). NO GUIDANCE IN SOURCE AGREEMENTS - Insurance minimums - Source code escrow Draft for attorney review and GC approval before use.
Tips
- •Use agreements of one type only. Mixing vendor and customer paper produces positions that contradict each other.
- •Include a few agreements where you gave ground. The playbook should reflect your realistic range, not only your best outcomes.
- •Answer every 'which to adopt' question Claude raises, then rerun with your answers added to the preferences block.
- •Once approved, save the playbook in a Claude Project and pair it with contract-red-flag-scanner or the triage workflow for incoming paper.
- •Have the responsible attorney approve the final playbook. Review it periodically as laws and your risk tolerance change.
Frequently Asked Questions
How many signed agreements do I need to build a useful playbook?
Three to ten agreements of the same type is a practical range. Fewer than three gives Claude too little to see a pattern, and the playbook becomes mostly your stated preferences. Many more can exceed what fits in one conversation. Choose a mix of recent deals, including some where you gave ground, so the fallbacks reflect what you actually accept.
How do I use the playbook once it is approved?
Save it as a document in a Claude Project for that contract type. When new paper arrives, paste it into a conversation in that Project and ask Claude to compare each clause to the playbook, labeling it standard, acceptable fallback, unacceptable or escalate. The workflow on triaging incoming contracts shows this step. Every comparison still goes to an attorney before it is acted on.
Should business teams see the playbook?
Many in-house teams share a simplified version with sales or procurement so they can accept standard and fallback positions without waiting on legal. If you do, remove internal commentary and keep the escalation triggers prominent. Decide internally who may approve fallbacks, and record that in the approval levels field so the playbook states it clearly.
Can the playbook keep up with changes in law?
Not on its own. The playbook reflects your past deals and your preferences at the time you build it. Assign an owner to review it on a regular schedule and whenever a relevant law or internal policy changes, particularly for data protection and limitation of liability positions. Rerun the prompt with newer signed agreements to see where your practice has drifted.
Related Prompts
Get New Prompts Like This Every Week
Join the free Claude for Lawyers newsletter — weekly prompts, tutorials, and practice-specific guides.