# Review methods and source boundaries ## Organization Preserve title, parties/preamble, recitals where useful, operative provisions, appropriate definitions, general provisions, signatures, and attachments according to the transaction. These are candidate sections, not mandatory boilerplate. Let the user retain an unconventional structure. Distinguish changing layout from changing meaning. Track moved passages by identity, not only text difference. ## Definitions Inventory definitions and their uses. Flag undefined, unused, circular, or conflicting definitions for review. Capitalization alone does not prove a defined term. Keep operative obligations out of definitions unless intentionally retained after review. Do not add a definition that broadens an obligation without identifying that effect. ## Contract concepts Distinguish factual assurances, future obligations, conditions, discretion, and declarations. Ask who must do what, by when, how, under which conditions, and with what consequence. Evaluate qualifiers from the represented party's perspective. Do not substitute shall/must/may mechanically without checking the intended legal function. ## Deal fidelity Compare each instruction against the draft using evidence-linked rows: expected term, actual text/location, match/mismatch/unknown, consequence, proposed correction. Check inclusive versus exclusive thresholds, deadlines, frequency, aggregate versus per-event limits, and whether negotiated permissions disappeared. A client narrative is evidence of intended terms, not proof the counterparty agreed. ## Termination Map trigger → notice → receipt/effective date → opportunity to cure → extension or plan → termination election → effective termination → surviving duties and accrued rights. Distinguish different cure periods and incurable breaches. Identify contradictory clocks or undefined triggers. Do not invent a remedy or declare enforceability from drafting logic alone. ## Verification Check semantic drift after restructuring; compare financial figures, dates, party roles, negation, exceptions, and conditions. Review linked provisions after edits. A clean mechanical report does not establish substantive completeness. ## Source provenance The installed Contract Navigator bundle was inventoried on 2026-10-01. Inspected materials included two numbering templates (one article-based), Chapter14SummaryChart.pdf (2 pages), Tips-for-Achieving-Clarity-in-Contract-Drafting.pdf (5 pages, Georgetown Writing Center, 2023), Exercise28-1CarrieRichards.docx, Exerxise31-8-ATerminationProvision.docx, and Last Class.docx. The 13-page 31-2.pdf contains additional exercises rather than the full Drafting Contracts textbook. Other indexed .doc and MindNode materials were found but not fully reviewed. Do not claim complete textbook coverage. The Carrie exercise juxtaposes a client narrative and contract; one instructional mismatch is an inclusive bonus threshold versus a strict greater-than threshold. Use that principle with original test fixtures, not copied source passages. The class notes flag obligations in definitions and inconsistent list/sentence presentation. The termination exercise supports explicit notice/cure/plan sequencing. These sources inform this original workflow; no source documents are bundled for resale. The user-supplied `CONTRACTS.zip` was reviewed on 2026-10-02. It contains Tina L. Stark's *Drafting Contracts* (2d ed.), the Georgetown Writing Center clarity handout, course notes and exercises, a sample internship agreement, and unrelated finance and constitutional materials. `transaction-review.md` synthesizes original prompts from the contract-related portions. The textbook, handout, notes, exercises, and sample agreement are not bundled. Several notes are compressed or potentially inaccurate statements of doctrine; verify relevant law and facts independently. Do not describe the skill as a licensed adaptation of the textbook. ## Cross-domain interpretive review Before proposing substantive changes, construct a compact relationship model: stated purpose; actors and roles; promised performances; resources and information each controls; timing; uncertainty allocation; decision rights; evidence of performance; and exit consequences. Separate express text, supplied context, inference, and unknowns. Validate every proposed correction against the actual transaction rather than transferring a familiar clause by superficial analogy. Apply four analytical lenses as question generators, not legal authorities: - Philosophical: distinguish description, promise, permission, condition, discretion, and acknowledgment. Ask what must be true for each obligation to arise and what counts as performance. Separate normative preference from the parties' actual agreement. - Linguistic: resolve actors, referents, scope of modifiers, quantifiers, inclusive/exclusive thresholds, conjunctions, exceptions, negation, and temporal language. Trace defined terms across provisions. Express alternative plausible readings before choosing one; ask for intent when consequential ambiguity cannot be resolved from evidence. - Psychological: examine the comprehension burden, salience of unusual obligations, and practical demands on attention and memory. Treat assumptions about incentives, biases, or likely behavior as hypotheses, never diagnoses or facts about these parties. Make decisions understandable without steering users into undisclosed risk allocation. - Relational/organizational: ask who can actually perform, authorize, observe, document, and enforce the operational steps. Consider information asymmetry, discretion, dependencies, repeated interactions, and continuity after personnel changes. Do not infer bargaining power or industry practice without support. For each material issue record: source text, lens, competing readings or practical failure scenario, supporting evidence, affected clauses, proposed change, whose interests change, uncertainty, and user decision needed. If the issue cannot be tied to contract language or supplied context, label it a question rather than a defect. Use a shared checklist before creating more agents. Where independent delegation is useful, one reviewer examines meaning and deal fidelity; another examines performance, incentives, and cross-clause effects. Both use the same evidence and decision record. Their findings remain proposals. Current controlling law and specific transaction facts constrain all analogies. Never present these lenses as a substitute for legal research or as validated predictions of human behavior.