sure-scale/doc-haus/contract-risk-checklist

Claude Code and agent skill by sure-scale/doc-haus

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

The canonical 12-area risk checklist for commercial contract review, with the red-flag patterns that mark each area as one-sided or dangerous. Use when reviewing any contract for risk — an NDA, MSA, SaaS agreement, services or consulting agreement, license, vendor terms, or order form — or when asked to spot issues, find red flags, assess a draft, mark up a counterparty's paper, or answer "what's wrong with this contract."

Install

npx vskill@latest install sure-scale/doc-haus/contract-risk-checklist

SKILL.md

Review every contract against all twelve areas below. Before you start, turn the twelve areas into an explicit checklist with the todowrite tool — one item per area — and answer each item present / absent / deviating rather than reviewing from memory, because a single skipped area can let an off-market term through. Mark an area completed only once you have located its governing clause (or confirmed it has none); an unresolved area stays in_progress. For each area: state what the contract says, cite the governing clause as [<Document> § <section>] (or note its absence — a missing clause is a finding, not a pass), and flag anything matching the red-flag patterns. Establish the governing-law clause first and qualify jurisdiction-sensitive findings against the matter's jurisdiction pack. For balanced market positions and fallback ladders per area, load the clause-library skill. A mapping of CUAD's 41-category clause taxonomy onto these twelve areas is in references/cuad-taxonomy.md.

<checklist>

1. Liability

Inspect: caps, exclusions, carve-outs, mutuality, super-caps for sensitive exposure.

What bad looks like:

  • Carve-outs from the cap so broad they swallow it — "any breach of this

Agreement," all indemnification obligations, or all confidentiality breaches uncapped in one direction.

  • One party's liability uncapped while the other's is capped.
  • A cap far below realistic exposure (one month's fees) with no separate

super-cap where the deal involves sensitive data or high-value IP.

2. Indemnification

Inspect: scope, triggers, who indemnifies whom, defense and settlement control, relationship to the liability cap.

What bad looks like:

  • One-way indemnity in a deal where both sides create risk.
  • Indemnity triggered by "any breach of this Agreement," converting every

contract claim into an uncapped indemnified claim.

  • Indemnitee controls defense and may settle at the indemnitor's cost without

the indemnitor's consent.

  • Indemnity covering first-party (direct) claims, not just third-party claims;

or no IP-infringement indemnity from the provider in a technology deal.

3. Termination

Inspect: for cause, for convenience, notice periods, cure rights, renewal mechanics, effect on fees and data.

What bad looks like:

  • Evergreen auto-renewal with a short or early opt-out window (non-renewal

notice due 90+ days before renewal) that is easy to miss.

  • One party may terminate for convenience but the other cannot; or cure

periods granted to only one side.

  • Termination accelerates all remaining committed fees regardless of cause.
  • No wind-down, transition assistance, or data-return obligation on exit.

4. Payment

Inspect: amounts, schedule, late fees, set-off, escalation, currency, taxes, refundability.

What bad looks like:

  • Unilateral price escalation — fees change on notice with no cap and no

corresponding termination right.

  • All fees non-refundable even where the provider is in breach; or payment

obligations that survive termination regardless of cause.

  • Set-off prohibited for one party while the other keeps broad set-off or

withholding rights.

  • All taxes and gross-up obligations shifted to one party.

5. Intellectual property

Inspect: ownership, license scope, work product, feedback and data rights, residuals.

What bad looks like:

  • Assignment of pre-existing IP, or of improvements to it, to the other side.
  • Customer-paid work product owned by the provider, or joint ownership with no

exploitation and accounting rules.

  • A license to customer data or feedback broader than needed to operate the

service (perpetual, irrevocable, "any purpose").

  • License grants that are silently perpetual or irrevocable where the

commercial intent is term-limited.

6. Confidentiality

Inspect: definition, duration, exclusions, permitted disclosures, return/destruction, adjacent restrictive covenants (non-solicit, non-compete, exclusivity).

What bad looks like:

  • One-way obligations where both parties disclose.
  • Marking-only definition with no reasonable-person catch-all; or a residuals

clause permitting use of anything retained in unaided memory.

  • Compelled-disclosure permission with no notice or minimization obligation.
  • Restrictive covenants smuggled in: no-hire (vs. no-solicit) provisions,

non-competes, or exclusivity beyond the deal's scope — enforceability is highly jurisdiction-dependent; qualify against the matter's pack.

7. Warranties

Inspect: express warranties, disclaimers, duration, remedies for breach.

What bad looks like:

  • "As is" disclaimer that reaches the core service commitment itself.
  • Warranty remedy that is illusory — sole remedy is re-performance at the

warrantor's discretion with no refund backstop.

  • One-sided compliance-with-laws or non-infringement warranties.
  • Disclaimers of implied warranties without the conspicuousness or scope

limits some jurisdictions require — qualify; do not assert a specific rule.

8. Limitation of remedies

Inspect: exclusive remedies, consequential-damages waivers, liquidated damages, who may enforce.

What bad looks like:

  • Consequential-damages waiver applied one-way.
  • Waiver listing "lost profits" without limiting it to consequential losses —

lost profits can be direct damages, so this quietly guts direct recovery.

  • An exclusive remedy that can fail of its essential purpose with nothing

behind it.

  • Liquidated damages that look punitive rather than a genuine pre-estimate of

loss — enforceability varies by jurisdiction; qualify.

9. Assignment and change of control

Inspect: consent requirements, deemed-assignment language, change-of-control triggers, affiliate and sale-of-business carve-outs.

What bad looks like:

  • Consent withholdable in sole discretion, combined with a deemed-assignment

clause making any change of control a breach — a hidden M&A blocker.

  • One side assigns freely while the other needs consent.
  • Termination right triggered by a change of control of one party only.
  • No carve-out for assignment to affiliates or in connection with a sale of

substantially all assets.

10. Governing law and dispute resolution

Inspect: governing law, venue, arbitration, jury waiver, fee shifting, equitable relief.

What bad looks like:

  • Counterparty home-court stack: their governing law, exclusive venue in their

forum, and a jury waiver together.

  • Mandatory arbitration with no carve-out to seek injunctive relief for

confidentiality or IP breaches.

  • One-way fee shifting (loser pays only when one side wins).
  • Shortened limitations periods or class waivers — enforceability varies by

jurisdiction; qualify.

11. Force majeure

Inspect: covered events, obligations during the event, notice and mitigation, exit rights.

What bad looks like:

  • Coverage of "economic hardship," market changes, or supplier cost increases

— turning ordinary business risk into an excuse for non-performance.

  • Only one party's performance excused.
  • No right to terminate after a prolonged event (commonly 30-90 days).
  • No notice or mitigation obligation on the party invoking it; or payment

obligations excused alongside performance.

12. Data protection

Inspect: controller/processor roles, security obligations, breach notice, subprocessors, international transfers.

What bad looks like:

  • Processing roles unassigned, or a processor granted rights to use personal

data for its own purposes.

  • Breach notification replaced by a long fixed window or omitted entirely.
  • No subprocessor flow-down, approval, or audit mechanism.
  • International transfers unaddressed where data will cross borders — the

required mechanism depends on the jurisdictions involved; check the matter's jurisdiction pack rather than assuming one regime.

</checklist>

These red-flag patterns are US commercial-market reference points, not legal standards; verify jurisdiction-sensitive items against the matter's jurisdiction pack before presenting them as findings.

<severity> Grade each area with one of four severities so the review reads as a triaged register, not a flat list. Use the same ladder for every area; the area's red-flag patterns above tell you what "off-market" looks like, this tells you how hard to push:

  • critical — off-market or missing in a way that exposes the client to

uncapped, unrecoverable, or deal-breaking risk (a cap-swallowing carve-out, a one-way uncapped indemnity, no IP-infringement indemnity in a technology deal). Must be fixed before signing.

  • high — a materially unfavorable term, or a protection a contract of this

type should always carry but does not (no termination-for-cause cure right, no data-return on exit). Fix, or record an explicit business decision to accept it.

  • medium — a negotiable deviation worth raising that the client can live with

if traded for something else.

  • low — cosmetic, market-standard with a minor wrinkle, or favorable to the

client; logged so coverage of the area is auditable.

When a term sits between two severities, take the higher one — an over-graded term costs the lawyer seconds to dismiss; an under-graded one goes unread into the signed agreement. An entirely-absent protection is graded on the exposure its absence creates, not downgraded for being a gap rather than bad language. </severity>

Security

  • Security scanPassed, 96 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/sure-scale/doc-haus/contract-risk-checklist/badge)](https://verified-skill.com/skills/sure-scale/doc-haus/contract-risk-checklist)

Something wrong with this skill or its page? Tell us.