deal-desk

Overview of deal-desk skill

What deal-desk does

deal-desk is a commercial review skill for evaluating a specific B2B deal before signature. It helps Revenue Operations, Deal Desk, sales leadership, finance, and legal teams turn a messy discount or redline request into a structured scorecard, margin analysis, contract-risk triage, and named human approval route.

The important design choice is that deal-desk never auto-approves. Its output is a decision aid: numeric scoring, risk flags, and a recommended approver chain such as Sales Director, VP Sales, CFO, CRO, or General Counsel.

Best fit for Revenue Operations teams

Use deal-desk for Revenue Operations when you need consistent handling of above-band discounts, non-standard payment terms, enterprise redlines, or deals with margin pressure. It is especially useful for SaaS and enterprise software teams where discounting, payment shape, DPA requirements, indemnity, liability caps, MFN clauses, and custom services work can materially change deal quality.

It is less useful for simple transactional sales, consumer pricing, or legal review of full contract prose without structured deal terms.

What makes this skill different

The skill includes three stdlib-only Python tools rather than just prompt instructions:

• scripts/deal_scorer.py scores a deal across margin, risk, strategic value, commercial policy, and term shape.
• scripts/discount_approval_router.py routes discount requests through policy bands and deal-size modifiers.
• scripts/terms_redliner.py detects structured contract landmines and recommends legal or commercial approvers.

The references also explain the underlying economics, including why a 30% discount on an 80% gross-margin product destroys 37.5% of margin dollars.

Adoption constraints to check first

Before installing, confirm that your team can provide structured intake data. The skill works best with ARR, TCV, list price, discount percentage, gross margin, payment terms, customer tier, strategic flags, and a normalized terms JSON. It is not a replacement for CRM governance, CPQ configuration, or counsel’s legal judgment.

How to Use deal-desk skill

deal-desk install and repository path

Install from the GitHub skill repository with:

npx skills add alirezarezvani/claude-skills --skill deal-desk

The source path is:

commercial/skills/deal-desk

After install, read these files first:

1. SKILL.md for the intended workflow and invocation boundaries.
2. assets/deal_intake_template.md for the required intake fields and JSON blocks.
3. references/discount_economics.md for margin math.
4. references/contract_landmines.md for terms-risk triggers.
5. scripts/ for executable scoring and routing behavior.

Inputs the skill needs

A strong deal-desk usage prompt should include the deal context, not just “review this discount.” At minimum, provide:

• Deal ID, customer name, owner, close date, segment, and customer tier.
• ARR, TCV, term length, list price, requested discount, and gross margin.
• Payment terms, billing shape, implementation cost, and custom work.
• Strategic justification: logo, reference, expansion, renewal, competitive displacement.
• Contract exceptions: indemnity cap, liability cap, DPA status, EU data, MFN, exclusivity, IP assignment, auto-renewal, governing law.

The intake template is valuable because its JSON blocks map directly to the scripts.

From rough request to complete prompt

Weak prompt:

Review this 32% discount for Acme.

Better prompt:

Use the deal-desk skill to review deal ACME-2026-Q2-117. ARR is $240,000, TCV is $480,000, term is 24 months, list price is $333,333, requested discount is 28%, product gross margin is 80%, payment terms are NET 60, customer tier is enterprise, and there is a written reference commitment. Terms include no indemnity cap, EU personal data, no DPA, MFN present, liability cap at 1x annual fees, and ambiguous IP assignment. Produce a scorecard, margin impact, redline list, and named approval route. Do not approve automatically.

This improves output because the skill can separate commercial flexibility from margin destruction and legal escalation.

Running the included tools

You can ask the agent to use the skill, or run the scripts directly for deterministic checks:

python scripts/deal_scorer.py --sample

python scripts/deal_scorer.py --input deal.json --profile saas

python scripts/discount_approval_router.py --input deal.json --output json

python scripts/terms_redliner.py --input deal_terms.json

Use script output as the factual base, then ask the AI to summarize tradeoffs, approval blockers, and next actions for RevOps, sales, finance, and legal.

deal-desk skill FAQ

Is deal-desk a legal review tool?

No. terms_redliner.py detects structured commercial-contract landmines such as uncapped indemnity, missing DPA for EU data, MFN clauses, problematic IP language, and unusual non-solicit or exclusivity terms. It is triage, not legal advice. High and critical findings should route to named counsel.

How is this better than an ordinary prompt?

A generic prompt may give plausible commentary but inconsistent routing. The deal-desk skill encodes policy-like behavior: scoring dimensions, discount bands, conservative escalation rules, and fixed-COGS margin math. That makes it better for repeatable deal review where consistency matters.

Can beginners use it?

Yes, if they start with assets/deal_intake_template.md. Beginners should avoid improvising input formats at first. Fill the template, generate the JSON blocks, run the sample scripts, and compare your real deal against the sample outputs before relying on the routing recommendation.

When should I not use deal-desk?

Do not use it to approve deals automatically, negotiate final legal language without counsel, evaluate consumer discounts, or scan full contract PDFs directly. For full prose contract review, use a dedicated contract scanner or legal workflow, then feed the structured exceptions back into deal-desk.

How to Improve deal-desk skill

Improve deal-desk inputs before changing logic

Most poor results come from incomplete intake. Add your actual price book assumptions, gross margin targets, approval names, discount authority bands, and customer segmentation rules. Replace placeholder approver roles with named owners if your operating model requires accountable routing.

For example, “VP Sales” is useful; “Jordan Lee, VP Sales, for enterprise deals over $500K ARR” is operationally better.

Calibrate policy bands to your business

The default router uses common discount bands: AE, Sales Manager, Director of Sales, VP Sales, then CFO/CRO. Your company may need different thresholds by product line, region, contract term, or customer tier.

If enterprise deals above $500K ARR always need VP review, encode that. If SMB renewals under $25K ARR should move faster, document the exception. The goal is consistent routing, not maximal escalation.

Watch common failure modes

Common mistakes include treating discount percentage as the whole story, ignoring implementation costs, failing to flag payment terms over NET 60, and under-routing contract risk because the composite score looks healthy. Critical terms should override a good commercial score.

Also avoid editing old intakes in place. The template recommends creating a new intake when pricing or terms change, which preserves approval history.

Iterate after the first output

After the first deal-desk output, ask targeted follow-ups:

• “Which two fields most changed the score?”
• “What concession would improve margin without changing ARR?”
• “What approval hop is triggered solely by discount versus contract terms?”
• “Rewrite the AE-facing summary in one paragraph with blockers and next steps.”
• “Show what score changes if payment moves from NET 75 to annual upfront.”

This turns the skill from a static review into a practical deal-desk guide for faster, better-governed close decisions.