--- name: contracts-basics-checks description: 15 rules from the Noesa course "Contracts, understood". For freelancers, founders, small-business operators, and working professionals who review ordinary service, vendor, contractor, or software agreements and use AI to summarize or draft them. For non-lawyers. --- # Contracts, understood — the rules Use with: Claude Code or Claude (save as a skill), Cursor (save under .cursor/rules as .mdc), ChatGPT or any other assistant (paste the text below into custom instructions or a project's instructions). 15 rules, taken from the course at https://noesa.leafsoft.online/c/contracts-basics Each heading is one thing the course teaches. Most are checks to run on your own output before presenting it as done; a few are background you are expected to have. 13 also name a mistake models make by default, under "Watch for". Apply these to the thing you are producing — the type, the schema, the query, the copy — not only to how you explain it. Where a rule names a field, a format or an identifier, that name belongs in the output. ## Read the deal before the clauses State a contract's business deal in four plain-language questions before reviewing its clauses. ## Identify who must do what Connect each important obligation to a named party, action, trigger, and deadline. ## Define scope without hidden work Test a scope for deliverables, exclusions, dependencies, and a change path. **Watch for:** If you ask AI to “make the scope comprehensive,” what durable mistake should you inspect for? AI may expand a common template into a long task list while lacking your actual boundaries, dependencies, and appetite for change. you must check whether the draft makes trade-offs visible rather than treating more words as more certainty. ## Pin down payment and acceptance Connect fees, invoices, payment dates, and acceptance to observable events. **Watch for:** If AI proposes “market-standard payment terms,” what should you challenge? AI defaults to common patterns without knowing your cash position, bargaining power, delivery sequence, or local rules. you must test the events and consequences against this deal rather than accept the label “standard.” ## Find ambiguous language Spot undefined, subjective, and conflicting language and turn it into focused review questions. **Watch for:** Why can an AI summary make an ambiguous contract appear clearer than the source? Summarization selects one plausible reading and often smooths over conflicts. Without your intended hours, thresholds, and document hierarchy, AI can convert uncertainty into false precision. you must compare the summary with the actual wording and preserve open questions. ## Understand warranties and promises Distinguish an operational promise, a warranty, and a remedy question in an ordinary agreement. **Watch for:** If AI summarizes a vendor's warranty as “strong protection,” what should you check before trusting that label? AI can read the assurance sentence while missing the disclaimers, exceptions, notice steps, and remedy limits placed elsewhere in the document. It cannot know your real exposure or whether the stated response matches the consequence the client would face. you must trace each promise to its evidence and remedy, and send broad, absolute, or contradictory language, and any question of legal effect, to a qualified lawyer. ## Trace liability when things fail Trace an ordinary failure from event to loss, responsibility, exclusion, and financial limit. **Watch for:** If AI says a liability clause is “balanced,” what missing work must you perform? AI may compare the wording with a common template while lacking your credible failure sizes, insurance, customer promises, and bargaining context. you must trace concrete consequences and ask qualified counsel about legal effect before accepting the label. ## Read indemnity as risk transfer Map an indemnity's trigger, covered claim, protected party, process, and limits. **Watch for:** Why might AI call an indemnity mutual while missing a one-sided transfer? AI can match mirrored headings or phrasing while overlooking different triggers, claim types, and exceptions. you must compare the consequence of each clause in your real failure scenarios, not its surface symmetry. ## Protect ownership and licenses Separate background materials, new work, ownership, and licensed permissions. **Watch for:** If AI drafts “all IP belongs to the client,” what hidden categories should you force it to separate? AI often reaches for a broad ownership pattern without knowing what pre-existed or came from third parties. you must separate background material, project work, dependencies, and the permissions the client needs in real operation. ## Handle confidentiality and data Map confidential information and personal data by source, permitted use, access, protection, and end-of-term handling. **Watch for:** Why can AI produce a detailed data clause that still misses your largest exposure? AI can fill a common clause pattern. It does not know your real data fields, customer promises, processing locations, subprocessors, or applicable regimes. you must supply and verify the data map. Then escalate jurisdiction-dependent duties. ## Plan termination before conflict Map termination triggers, notice, cure, transition duties, and surviving obligations. **Watch for:** If AI finds a termination-for-convenience clause, why should it not report “the customer can leave freely”? AI may compress the right to give notice into a complete exit while missing fees, minimum terms, export limits, survival, and operational dependency. you must simulate the whole departure and preserve uncertainty about legal effect. ## Notice renewal and lock-in Calculate a renewal decision date and identify price, notice, and minimum-term constraints. **Watch for:** Why might an AI summary state the term length correctly yet miss the commercial commitment? AI may extract the prominent twelve-month phrase while failing to combine renewal, notice, pricing, minimum spend, and exit mechanics across sections. you must calculate the lifecycle as one system. ## Resolve conflicts and change terms Map escalation, dispute handling, document precedence, and amendment authority. **Watch for:** An AI drafts a neat disputes-and-changes clause. What does it not know about the country whose law governs the contract? AI can assemble a conventional escalation ladder and change-control clause while missing which document actually controls in your real set, who holds authority to bind each party, and whether a casual message could operate as an amendment. It cannot judge the local effect of a chosen forum, governing law, or waiver. you must map the actual document set and authority, and route choice of forum, governing law, and amendment language to a qualified lawyer when the consequences matter. ## Challenge an AI-drafted contract Audit an AI-drafted agreement against the intended deal, missing context, internal conflicts, and failure scenarios. **Watch for:** Why can asking AI for a confidence score make the review worse? A numerical score can create false precision without an agreed standard or complete context. you needs traceable clauses, explicit assumptions, failure scenarios, and named uncertainties, not a number that disguises what the system cannot know. ## Decide when to sign or escalate Turn a completed contract map into a reasoned proceed, renegotiate, pause, or legal-review decision. **Watch for:** What should you ask AI for at the decision point instead of “Should I sign?” Ask for a clause-linked list of assumptions, unresolved questions, competing readings, and failure scenarios. AI can organize evidence, but it cannot know your full context, risk authority, or local legal effect, so the accountable human and qualified advisers make the decision.