ooligo
prompt

NDA clause redline prompt pack

Difficulty
beginner
Setup time
15-30 min
For
legal-ops-manager
Legal Ops

Stack

A pack of twenty-four structured Claude prompts for redlining the twelve most-negotiated NDA clauses. Each clause gets two prompts: an A-prompt for the initial redline of the counterparty’s draft against your firm’s playbook, and a B-prompt for drafting a counter-position when the counterparty pushes back. The prompts are calibrated to anchor against the firm’s own NDA playbook — without the playbook, the pack doesn’t invent positions, it flags the gap. Every output is a draft the contracts manager or in-house counsel reviews and edits before sending. Consult counsel on any clause position before committing to it.

The twelve clauses: confidentiality scope, definition of Confidential Information, exclusions from Confidential Information, permitted disclosures, return or destruction, residuals, non-solicitation, IP ownership, governing law, injunctive relief, NDA term, and survival of confidentiality obligations.

When to use

  • The firm has a written NDA playbook with named positions per clause (preferred, acceptable, walk-away). Without the playbook, the prompts flag every clause as “playbook does not cover” rather than asserting positions.
  • NDA volume is high enough that consistency across negotiations matters — typically 10 or more NDA reviews per quarter.
  • A contracts manager or paralegal is doing first-pass review and escalating flagged items to in-house counsel, rather than counsel reviewing every NDA from scratch.
  • You want the review to consistently catch informal-channel and personal-account issues — these prompts ask about residuals, non-solicitation, and IP in an NDA structure that often introduces provisions a standard review misses.

When NOT to use

  • Highly bespoke or novel NDA structures. Joint-development NDAs, NDAs bundling exclusivity obligations, or NDAs tied to M&A processes carry strategic dimensions the prompts don’t cover. Those need counsel from scratch.
  • Replacing counsel’s review on high-value deals. The pack is a first-pass tool for volume NDA work. When the deal at the other end of the NDA is material — a significant customer, an M&A counterparty, a strategic partnership — treat the pack output as a starting framework and have counsel complete the review.
  • Auto-sending redlines. The pack produces drafts; the contracts manager sends. No output from this pack goes directly to a counterparty without human review and sign-off.
  • Non-commercial NDAs. Employee NDAs, NDA components embedded in employment agreements, and NDAs in regulated industries (healthcare, financial services) carry compliance dimensions beyond commercial negotiation. Consult counsel on the appropriate review process.

Setup

  1. Download the bundle. Place apps/web/public/artifacts/nda-clause-redline-prompt-pack/nda-clause-redline-prompt-pack.md in a location your contracts team can read — a Notion page, an internal wiki, or a Claude project’s knowledge files.
  2. Author the firm’s NDA playbook. Per the twelve clauses in the pack, document: preferred position, acceptable position, walk-away position, rationale for each level. The playbook is the anchor the prompts cite; without it, the prompts flag gaps rather than generating positions. Authoring the playbook is a one-time project, typically 8-15 hours of counsel time, that pays back across every subsequent NDA review.
  3. Create a Claude project per active negotiation. Name it nda-<counterparty>-<matter-id>. Drop the playbook and the counterparty’s current draft as project knowledge. Add the pack’s prompts as saved prompts within the project.
  4. Run A-prompts for first-pass review. Prompts 1A through 12A cover the initial redline of the counterparty’s draft. Run them in sequence on the first draft you receive.
  5. Run B-prompts after counterparty responds. For each clause where the counterparty pushed back on your redlines, run the corresponding B-prompt. The B-prompts draft a counter-position anchored in the playbook.
  6. Review and edit before sending. The contracts manager owns voice, judgment, and the final text. Consult counsel on any clause before committing to a position.

What the pack contains

Twenty-four prompts across twelve clauses, organized as paired A/B prompts.

Tier 1 — Confidentiality framework (Clauses 1-4). Prompts 1A/1B cover confidentiality scope (mutual vs one-way, standard exclusions). Prompts 2A/2B cover the definition of Confidential Information (category coverage, oral disclosure confirmation windows). Prompts 3A/3B cover exclusions (the four standard carveouts — public domain, prior knowledge, independent development, third-party disclosure). Prompts 4A/4B cover permitted disclosures (need-to-know employees, professional advisors, compelled disclosure with advance notice).

Tier 2 — Obligations and restrictions (Clauses 5-8). Prompts 5A/5B cover return or destruction (certification timing, backup-system exceptions with ongoing obligation). Prompts 6A/6B cover residuals (narrow unaided-memory scope vs overbroad “intangible residual information” language — a distinction that has been litigated in trade-secret cases). Prompts 7A/7B cover non-solicitation (mutual vs one-way, passive-recruitment exclusion, jurisdiction sensitivity). Prompts 8A/8B cover IP ownership in an NDA (the key failure mode: IP assignment appearing in an NDA context where it doesn’t belong; the prompts flag this as out-of-scope and escalate to counsel).

Tier 3 — Enforcement mechanics (Clauses 9-12). Prompts 9A/9B cover governing law (jurisdiction, conflicts-of-law rule, downstream enforceability implications for non-solicitation and residuals). Prompts 10A/10B cover injunctive relief (irreparable-harm acknowledgment, no-bond provision, mutuality). Prompts 11A/11B cover NDA term (duration, termination mechanism, auto-renewal). Prompts 12A/12B cover survival of confidentiality obligations (general CI survival period, trade-secret carveout in perpetuity, the DTSA framing for US-governed NDAs).

Each prompt is structured: role / context / input format / task / things-to-avoid (the specific failure mode the guard targets) / output format. The “things-to-avoid” sections name the actual failure modes — accepting one-way non-solicitation without flagging it, agreeing to IP assignment in an NDA context, dropping the advance-notice requirement on compelled disclosures — not generic hedging.

Cost reality

  • Per-prompt token cost — each prompt invocation uses roughly 8,000-20,000 input tokens (playbook + counterparty draft + prompt instructions) and 500-1,500 output tokens. Per prompt: $0.05-0.10 at current Claude pricing ($3/M input, ~$15/M output).
  • Per-NDA review, full first-pass (12 A-prompts): ~$0.60-1.20.
  • Per-NDA review, full lifecycle including B-prompts for disputed clauses (typically 4-6 of the 12 get pushed back): ~$1.50-2.50.
  • Contracts manager / paralegal time — the time saving is the primary ROI. A manual first-pass review of a counterparty NDA against a playbook takes 60-90 minutes for a paralegal and 30-45 minutes for an experienced contracts manager. With the A-prompts plus review of output: 15-25 minutes per NDA. At 40 NDA reviews per quarter, that’s 30-40 hours saved per quarter.
  • Setup time — 15-30 minutes once the playbook exists. Playbook authoring is the binding one-time cost (8-15 hours of counsel time); it is reused across every NDA, vendor agreement with NDA attachments, and M&A confidentiality agreement.

Spellbook integrates NDA review directly into Word and can apply playbook-anchored redlines in-flow. If your contracts team lives entirely in Word and wants in-document redlines rather than Claude outputs they copy into Word, Spellbook’s in-document experience is faster. The trade-off: Spellbook requires a license (pricing starts at a few hundred dollars per seat per month, estimate based on published positioning — verify current pricing) and is limited to its built-in clause coverage. This prompt pack has no per-seat licensing cost and covers clause logic the contracts manager can inspect and modify.

Success metric

  • Time from counterparty draft received to firm redlines sent. Should drop from 3-5 business days to 1-2 business days for standard NDA reviews.
  • Clause-escalation rate. Share of clauses per NDA that require counsel review beyond the contracts manager’s sign-off. Should settle at 1-3 clauses per NDA after the first quarter of use — if it’s higher, the playbook walk-away thresholds need tightening; if it’s lower, the escalation criteria may be too narrow and the contracts manager may be signing off on positions that warrant counsel eyes.
  • Playbook-citation accuracy. At quarterly review, pull 10 recent NDA redlines and verify that every position the pack cited from the playbook actually appears in the playbook at the cited level (preferred/acceptable/walk-away). Drift between the pack’s citations and the current playbook means the playbook was updated without updating the project knowledge files.

vs alternatives

vs Spellbook or Harvey NDA review modules. Those tools apply AI redlining inside Word with firm-playbook integration. Pick them when the contracts team’s primary interface is Word and the per-seat license is within budget. This pack is the right choice when you want full visibility into the prompt logic (the contracts manager can read and edit every prompt), when you need to cover clause logic those tools don’t expose, or when you’re doing low-volume NDA work that doesn’t justify a per-seat license.

vs ChatGPT-style “redline this NDA against this playbook.” A generic chat prompt returns paragraph-level commentary. These prompts are structurally different: per-clause output with playbook tier cited, specific replacement language, explicit things-to-avoid sections that guard specific failure modes (one-way non-solicitation accepted without flag; IP assignment not caught; residuals scoped too broadly). The structural difference is the reason to use this pack rather than a generic prompt.

vs manual paralegal review. Right for the highest-stakes NDAs where counsel’s judgment on strategic trade-offs dominates every clause. The pack earns its setup cost on the high-volume commercial NDA work — vendor onboarding NDAs, customer NDAs before deal qualification, partner NDAs before due diligence — where paralegal bandwidth is the constraint and consistency across negotiations is the goal.

Watch-outs

  • Playbook drift. The prompts cite playbook positions explicitly; if the playbook is updated (positions change, new clauses are added) without updating the Claude project knowledge, the prompts produce stale citations. Guard: prompts flag “playbook does not cover” when a clause isn’t in the playbook rather than inventing a position. Stale playbook coverage shows up as “does not cover” on clauses the firm actually has positions on — a visible signal that the project knowledge needs updating.
  • Jurisdiction-specific enforceability. Non-solicitation, residuals, and governing law have jurisdiction-specific enforceability implications the prompts flag but cannot resolve. Guard: prompts for non-solicitation (7A/7B), governing law (9A/9B), and residuals (6A/6B) include explicit counsel-escalation flags for jurisdiction-sensitive positions. The contracts manager should not resolve these flags without counsel input.
  • IP assignment masquerading as an NDA clause. Counterparties occasionally insert IP-assignment language into NDA templates — “any improvements or derivatives based on the Confidential Information shall belong to the Disclosing Party.” This is outside the scope of what a contracts manager should negotiate in an NDA. Guard: prompt 8A explicitly flags this pattern and escalates to counsel rather than generating a counter-redline. The contracts manager does not resolve IP-assignment questions without counsel sign-off.
  • Auto-send drift. The pack produces drafts. Guard: none of the prompts produce output formatted as a ready-to-send email or cover letter. The output is always a redline draft or analysis, requiring explicit contracts manager action to convert to a sending-ready format.

Stack

The bundle lives at apps/web/public/artifacts/nda-clause-redline-prompt-pack/:

  • nda-clause-redline-prompt-pack.md — the twenty-four prompts, paste-ready

Tools: Claude for the prompts; Spellbook as the named in-Word alternative if your team prefers in-document redlining. The output drops into Word redline format, your CLM, or back into the negotiation thread.

Related: the MSA negotiation prompt pack covers a broader MSA negotiation workflow including fallback laddering and counterparty-posture reading.

Files in this artifact

Download all (.zip)