Operationsskill

NDA review-redline-sign skill — built on Claude's default NDA skill, ships signed PDF

Problem solved: Founders and fundraising ops drown in incoming NDAs; a custom Claude skill (on top of Anthropic's default NDA-review skill) analyzes each NDA, generates a redline .docx with tracked changes per company standards, and outputs a clean PDF with the founder's signature inserted.

TierBMedium — Solid coverage with concrete examples
EffortM · Weeks
Forfounder · BD lead · fundraising ops · COO

What

The operator paste-loads an inbound NDA file into Claude Code. A custom skill runs the analysis pass (using Anthropic's bundled NDA-review skill as the base), proposes redlines according to a per-company standards doc, generates a .docx with tracked changes, generates a clean PDF, and inserts a stored signature image into the signature block. The operator confirms which redlines to keep, then ships.

Why it matters

A fundraising founder reported signing 6 NDAs in one week using this flow, with counterparty turnaround in 1-2 days. Two effects beyond raw speed: (1) signals to counterparties that the company can move quickly, (2) the redlines are reasoned, not boilerplate, so they pass legal review on the other side.

End-to-end

  1. Start from Anthropic's default NDA-review skill (bundled in Claude). Don't rebuild what's already there.
  2. Author a company-specific standards doc: which clauses you accept, which you redline, your standard limitations of liability, term, governing law, mutual-vs-one-way preferences.
  3. Build a skill that takes an inbound NDA, runs the default NDA review for analysis, then applies your standards doc on top to produce the redline.
  4. Have the skill emit a .docx with tracked changes (so counterparty sees what changed) plus a clean PDF with your signature inserted.
  5. Explicitly instruct the skill not to sign correspondence as "Claude" — easy to forget; awkward when caught.
  6. Operator confirms each redline in chat before the skill finalizes the file.
  7. Send. Track win-rate (how often the counterparty accepts your redlines unchanged) to refine the standards doc.

Gotchas

The skill will, if not constrained, sign its own correspondence as "Claude" in cover messages — caught in production, embarrassing. Add an explicit "never sign as Claude; use the configured human name" rule. Second gotcha: standards docs are company-vertical and counterparty-dependent (an enterprise vendor's standards differ wildly from a fund's). Expect to maintain 2-3 standards docs, not one.

Tools

  • Claude Code with skill support.
  • Anthropic's default NDA-review skill as the foundation.
  • A company-specific standards doc (the actual content of redlines; this is the company-vertical piece — write it once, refine over months).
  • A stored signature image; PDF generation in the skill.
← All casesB-111