Comparing Two Contract Versions in 30 Seconds — Every Change Highlighted

The Problem

Here's a scenario every founder knows. You send a term sheet. The investor sends back "v2 with minor edits." You open both documents. You squint. You try to remember what the original said. You run a text diff. It highlights 47 changes including reformatted paragraphs, moved commas, and renumbered sections.

Buried in there? Three changes that fundamentally shifted economics and two that removed your protections. But they look identical to the 42 cosmetic edits.

Text diffs are useless for contracts. They tell you what changed. They don't tell you why it matters.

When we were negotiating a strategic partnership for one of our portfolio companies, the counterparty's lawyers sent back a "lightly revised" agreement. It had 31 tracked changes. Our lawyer needed 4 hours to work through them. By the time she was done, the negotiation window had narrowed. We signed two days later than we should have.

I hated that. Two days lost because diffing contracts is manual labor.

The Solution

The Contract Analyzer skill includes a semantic comparison mode. Upload two versions of the same contract. The agent doesn't just find text differences — it understands what each change does. Who gains. Who loses. What protections were added or removed.

Color-coded. Annotated. In 30 seconds.

The Process

skill: contract-analyzer
input:
  mode: compare
  original: /contracts/term-sheet-v1.pdf
  revised: /contracts/term-sheet-v2.pdf
  perspective: "company"  # analyze changes from our perspective
output:
  format: markdown
  include_unchanged: false  # only show what moved

The agent performs three passes:

Pass 1 — Structural Alignment. Maps sections between versions even if numbering changed. §4.2 in v1 might be §5.1 in v2 — the agent tracks it.

Pass 2 — Semantic Diff. For each changed clause, the agent classifies:

🟢 Favorable  — Change benefits your position
🔴 Unfavorable — Change weakens your position
🟡 Neutral     — Different but materially equivalent
⚪ Cosmetic    — Formatting, numbering, typos

Pass 3 — Impact Analysis. Each material change gets a brief:

## Semantic Diff: Term Sheet v1 → v2

**Material Changes: 5 of 31 total edits**
**Net Direction: Unfavorable (3 red, 1 green, 1 yellow)**

### 🔴 §3.1 Liquidation Preference
- v1: "1x non-participating preferred"
- v2: "1.5x participating preferred"
- **Impact:** Significant. Changes from standard to aggressive.
  Participating preferred + higher multiple = founders receive
  materially less in most exit scenarios below €50M.

### 🔴 §6.2 Board Composition
- v1: "2 founder seats, 1 investor seat, 1 independent"
- v2: "2 founder seats, 2 investor seats, 1 independent"
- **Impact:** Board control shifts. Investor bloc can now match
  founder votes and sway the independent.

### 🟢 §8.1 Anti-Dilution
- v1: "Full ratchet anti-dilution"
- v2: "Broad-based weighted average anti-dilution"
- **Impact:** Favorable. Weighted average is founder-friendly.
  Reduces dilution impact in down rounds.

The Results

The real metric? Decision speed. When you can see in 30 seconds that v2 is materially worse, you call the investor that afternoon instead of waiting for your lawyer next Tuesday.

Try It Yourself

# Install via Mr.Chief dashboard after signing up at mrchief.ai/setup
# clawhub install contract-analyzer

Then:

Compare these two contract versions. Show me what changed, who benefits from each change, and flag anything that weakens our position.

Upload both files. Get your answer before your coffee gets cold.

Thirty-one edits. Five that mattered. Three that were traps. Found in thirty seconds.