本文へ移動
cccskills
無料GitHub で公開

deal-closer-playbook

Analyzes a deal in progress and generates a comprehensive closing strategy. Researches the target company, maps the buying committee, builds objection responses, creates competitive positioning, and outputs a tactical deal-playbook.md with next-best-actions and a mutual close plan.

インストール方法を見る

含まれるファイル(4)

  • SKILL.md4.7 KB
  • references/intelligence-gathering.md5.5 KB
  • references/output-template.md3.6 KB
  • references/playbook-strategy.md6.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Deal Closer Playbook

Take a deal in progress -- at any stage from discovery to negotiation -- and produce a tactical closing playbook combining company research, stakeholder mapping, competitive intelligence, and deal mechanics into a single actionable document.

Contents

  • references/intelligence-gathering.md -- context intake, company research, buying-committee map, MEDDIC and velocity risk scoring
  • references/playbook-strategy.md -- objection matrix, competitive positioning, stage-specific closing strategy, mutual close plan, proposal talking points
  • references/output-template.md -- full deal-playbook.md structure to write

Workflow

Operate in two phases: gather intelligence, then generate the playbook. Be thorough but fast. Tie every section to a specific action.

  1. Collect deal context. Gather the required, valuable, and nice-to-have inputs. Ask for anything missing; mark unavailable items [UNKNOWN] and work around them. See references/intelligence-gathering.md.
  2. Research the company. Use WebSearch for current intelligence: overview, last-90-days news, financial signals, leadership/hiring, tech-stack signals, industry context. See references/intelligence-gathering.md.
  3. Map the buying committee. Identify and profile each role (champion, economic buyer, technical evaluator, user buyer, coach, blocker, procurement/legal, executive sponsor). Flag unknown stakeholders as discovery gaps. See references/intelligence-gathering.md.
  4. Assess deal risk. Score MEDDIC qualification and the velocity risk checklist. See references/intelligence-gathering.md.
  5. Build the objection response matrix. Address every raised objection plus likely unraised ones for the stage and context. See references/playbook-strategy.md.
  6. Build competitive positioning. For each competitor (or the status quo), document their pitch, where they win, where you win, landmine questions, and traps to avoid. See references/playbook-strategy.md.
  7. Design the stage-appropriate closing strategy. Match tactics to discovery/demo, evaluation/proposal, or negotiation/close. See references/playbook-strategy.md.
  8. Build the mutual close plan. Create the shared buyer-seller timeline to signed contract. See references/playbook-strategy.md.
  9. Generate proposal talking points. Draft the opening, value prop, proof, differentiation, and the ask. See references/playbook-strategy.md.
  10. Write the deal playbook. Output the complete document to deal-playbook.md using references/output-template.md.

Behavioral Rules

  1. Be honest about deal risk. If the deal is poorly qualified, say so -- false confidence is worse than a clear-eyed adjustment.
  2. Research before advising. Do not build competitive positioning from general knowledge alone; use WebSearch for current information.
  3. Make it actionable. Every section answers "What do I do next?" If a section leads to no action, cut it.
  4. Write for the rep, not the VP. Use direct, tactical language. This is a field guide, not a board deck.
  5. Be specific about timing. "Send the ROI model to Sarah by Thursday at 2pm before her Friday leadership meeting" beats "follow up soon."
  6. Do not manufacture information. Mark anything not found as a gap and recommend how the rep can get it.
  7. Respect the competitive landscape. Highlight genuine differentiators; let the prospect draw conclusions rather than trashing competitors.
  8. Think like the buyer. Every recommendation must pass "Would this make me more or less likely to buy?" Pushy tactics and artificial urgency destroy trust.

Edge Cases

  • No competitors identified: Build a competitive section against the status quo (doing nothing, building in-house, manual processes). Every deal competes against inaction.
  • Very early stage deal: Focus on qualification and discovery; the closing strategy becomes an advancement strategy. Do not write negotiation tactics for a pre-demo deal.
  • Stalled deal: Diagnose why it stalled (timing, budget, champion left, competing priority) and build a re-engagement strategy around a new compelling event.
  • Renewal/expansion deal: Shift from acquisition to retention framing. Emphasize value delivered, usage data, and ROI proof over competitive positioning.
  • Multi-product deal: Map each product to a different stakeholder and use case. Build separate value propositions per product line under a unified closing strategy.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Audit websites for accessibility issues and WCAG compliance. Use when checking accessibility, fixing a11y issues, or ensuring WCAG compliance.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

Deploy a 2-layer parallel agent hierarchy for large, parallelizable work — big refactors, multi-file migrations, codebase-wide audits, bulk generation. A top-tier commander (Fable or Opus) orchestrates the swarms; the user picks a power level (Max Power / Heavy / Balanced / Economy) that sets the Opus/Sonnet/Haiku model mix per layer. Layer 1 is 3-50+ specialist agents, each with its own full context window; Layer 2 is 2+ sub-agents per member. Includes git safety, tiered sizing, a pre-deploy gate, phantom-completion checks, and multi-wave follow-up.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

Use to test whether an AI agent, prompt, or skill actually works before or after it ships — build a small eval set, score runs, and catch regressions when the prompt or model changes. Trigger when the user asks whether a prompt is good, wants to compare models or versions, sees inconsistent agent output, or is about to put an agent in front of customers.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

Deploys swarms of sub-agents for massive parallel data processing tasks. Unlike agent-army (which is for code changes), this is for DATA tasks -- processing 1000 documents, analyzing datasets, bulk content generation. Configurable swarm size, task distribution, result aggregation, progress tracking, and error recovery.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

Designs and deploys custom agent teams for specific business workflows. Interactive discovery of business processes, then generates complete team configurations with specialized agent roles, tool access, communication protocols, and handoff rules.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

Agent-to-Agent (A2A) communication protocol. Connect two or more Claude agents that pass messages, share context, delegate tasks, and collaborate. Implements structured handoffs, shared memory, and multi-agent conversations.

日本語の概要は準備中です。原文の説明を表示しています。

OneWave-AI/claude-skills3392026年10月2日 更新

OneWave-AI のスキルをすべて見る

このスキルの問題を報告する