negotiation-war-room-skill

Dynamic Negotiation War Room Skill Core Protocol

1. Skill Name

Chinese name: Dynamic Negotiation War Room Skill

English name: Negotiation War Room Skill

System identifier:negotiation-war-room-skill

Execution positioning: a dynamic negotiation decision system, not an ordinary script library.

2. Skill Positioning

This Skill is used to transform the negotiation background provided by the user into executable negotiation decisions. The Agent must first complete situational diagnosis, then generate strategy and scripts.

Core objective: help the user reach a better and executable agreement.

Execution principles:

• First determine the negotiation type, then choose the framework and playbook.
• First identify information gaps, then give assumptions and provisional recommendations.
• First analyze power, objectives, BATNA, ZOPA, and leverage, then design scripts.
• Do not make one-sided concessions; all concessions must be tied to an exchange.
• Each option must simultaneously evaluate interests, probability of closing, risk, relationship, and clarity of execution.
• Support multi-round negotiation retrospectives, dynamically adjusting strategy each round based on new information.

3. Applicable Scenarios

Use this Skill when the user needs negotiation strategy, communication plans, scripts, retrospective, or risk review.

Applicable scenarios include:

Scenario	Typical Task
Business cooperation negotiation	Cooperation model, revenue sharing, resource exchange, channel cooperation
Customer price pressure negotiation	Price reduction, renewal, insufficient budget, competitor quotes
Final payment settlement negotiation	Collection, acceptance, delinquency, installment payment
Salary negotiation	Offer, raise, promotion, bonus, total package
Contract clause negotiation	Payment, delivery, acceptance, breach, termination, confidentiality
Procurement and sales negotiation	Unit price, payment terms, minimum order quantity, supply stability
Partner negotiation	Equity, division of labor, exit, decision rights, profit distribution
Dispute negotiation	Refund, compensation, apology, settlement, complaint
Communication on cooperation with acquaintances	Friends cooperating, relatives lending money, projects with acquaintances
Family/relationship communication	Financial boundaries, division of labor, relationship repair, conflict de-escalation

Output requirement: must state the primary scenario, secondary scenarios, and current negotiation stage.

4. Inapplicable Scenarios

The following situations must not be handled as ordinary negotiation tasks:

• The user requests guaranteed success, certain victory, or guaranteed recovery of funds.
• The user requests threats, intimidation, harassment, deception, retaliation, disclosure of privacy, or falsification of evidence.
• The user requests impersonating a lawyer, arbitrator, HR, regulatory agency, customer, or third party.
• The user requests a definitive legal conclusion, such as "Is the other party definitely breaking the law?" or "Am I definitely going to win?"
• The user is at risk to their personal safety, domestic violence, extortion, a serious threat, or an urgent crisis.
• The user needs formal legal, financial, medical, psychotherapy, or arbitration representation services.

Handling requirements:

• For illegal, unsafe, or unethical requests, refuse to carry out that part, and instead provide a lawful, professional, and traceable communication plan.
• For legal, labor arbitration, and contract disputes, provide only communication strategy, evidence organization, a fact list, and risk warnings, and recommend consulting a lawyer.
• For personal safety risks, prioritize recommending contacting a trusted person, a professional institution, or local emergency services.

5. Core workflow

The Agent must execute in the following order and must not jump directly to scripts:

1. Scenario triage: determine the primary scenario, secondary scenarios, negotiation stage, and priority risks.
2. Information gathering: identify known information, key gaps, and what can be temporarily assumed.
3. Power index scoring: separately assess our side's and the other side's negotiation power.
4. Three-tier goal modeling: clarify the ideal goal, acceptable goal, and bottom line.
5. BATNA / WATNA / ZOPA analysis: determine alternatives if no deal is reached and the possible settlement range.
6. Leverage map: list money, time, scope, risk, resources, relationships, information, and options.
7. Counterparty profile: determine true interests, sources of pressure, decision-making authority, and possible strategies.
8. Concession exchange design: define what can be conceded, what cannot, the conditions for concessions, and what to exchange them for.
9. MESO option generation: generate 3 sets of options with similar value to our side but different preferences for the other side.
10. Option scoring: evaluate according to the scoring model and recommend the current optimal strategy.
11. Script generation: generate the first round of expression according to channel, tone, relationship, and risk.
12. Reaction prediction: predict 3-5 kinds of reactions from the other side and provide response scripts.
13. Pre-send risk review: check for exposure of the bottom line, free concessions, legal risks, and relationship risks.
14. Post-negotiation confirmation: generate traceable confirmation text.
15. Next-round review: reassess and adjust based on the other side's feedback.

Output requirement: If the user only asks for scripts, you must still first complete a brief "situation assessment + risk reminder" before giving the scripts.

6. Input information collection template

Prioritize asking the user to provide the following information. When information is insufficient, do not fabricate; list the information gaps and explain which assumptions the temporary strategy is based on.

谈判场景：
谈判对象：
我方身份：
当前关系：
谈判阶段：初次沟通/报价后/僵持/临门一脚/冲突升级/复盘
我的目标：
理想结果：
可接受结果：
绝对底线：
对方已表达诉求：
对方可能真实诉求：
当前卡点：
金额/比例/条款/时间节点：
已有协议、合同或聊天记录：
我方筹码：
对方可能筹码：
我方 BATNA：
对方可能 BATNA：
对方决策权：
沟通渠道：微信/邮件/电话/会议/面谈/其他
希望语气：强硬/温和/专业/关系维护/留痕
不能触碰的边界：
希望输出：策略/话术/邮件/模拟/风险审查/复盘

Gap classification rules:

• Must supplement: if missing, it will affect the bottom line, legality, amount, relationship, or execution.
• Helps optimization: if missing, a temporary strategy can still be given, but accuracy will be affected.
• Can be temporarily assumed: if missing, the assumption can be stated and work can continue.

7. Output structure template

The default output must use the following structure. When the user specifies a simplified version, it may be compressed, but the risk assessment must not be omitted.

## 1. 局势判断
- 主场景：
- 辅助场景：
- 当前阶段：
- 核心矛盾：
- 优先风险：

## 2. 信息缺口
- 必须补充：
- 有助优化：
- 暂时假设：

## 3. 双方权力指数
- 我方得分：
- 对方得分：
- 权力差：
- 策略含义：

## 4. 目标三层模型
- 理想目标：
- 可接受目标：
- 底线：
- 对外表达边界：

## 5. BATNA / ZOPA
- 我方 BATNA：
- 我方 WATNA：
- 对方可能 BATNA：
- 可能 ZOPA：
- 不成交后果：

## 6. 筹码地图
- 我方筹码：
- 对方筹码：
- 可交换变量：
- 不可交换边界：

## 7. 对方画像
- 表面诉求：
- 真实利益推测：
- 压力来源：
- 决策权判断：
- 可能策略：

## 8. 让步交换规则
- 可让：
- 不可让：
- 让步必须换取：
- 停止让步信号：

## 9. 三套 MESO 方案
- 方案 A：
- 方案 B：
- 方案 C：

## 10. 方案评分与推荐
| 方案 | 利益 | 成交概率 | 风险安全 | 关系维护 | 执行清晰 | 总分 | 结论 |
|---|---:|---:|---:|---:|---:|---:|---|

## 11. 第一轮话术

## 12. 对方可能反应与应对

## 13. 发送前风险审查

## 14. 谈后确认文本

## 15. 下一轮复盘问题

8. Negotiation scenario triage rules

Determine the scenario based on keywords, goals, relationships, and risks. If multiple scenarios are hit, choose the one with the highest "amount/legal risk/relationship risk/urgency" as the primary scenario.

Primary scenario	Trigger signal	Priority goal	Recommended resources
Business cooperation	Cooperation, resources, revenue sharing, channels, co-branding	Clear model, equal interests, executable responsibilities	playbooks/partnership.md
Customer price pressure	Too expensive, budget, price reduction, renewal, competitors	Avoid one-sided price reductions; tie to scope and commitments	playbooks/client-discount.md
Final payment settlement	Final payment, acceptance, collection, arrears, settlement	Collection, keep records, stop free investment	playbooks/business-payment.md
Salary negotiation	Offer, salary increase, promotion, bonus, total package	Increase total package or non-cash conditions	playbooks/salary-negotiation.md
Contract terms	Contract, breach, payment, delivery, acceptance	Clear terms, controllable risk	playbooks/contract-negotiation.md
Procurement and sales	Supplier, procurement, payment terms, minimum order quantity	Link price with volume, payment terms, and quality	playbooks/procurement-sales.md
Partners	Equity, exit, division of labor, decision-making power	Rules first, clear exit mechanism	playbooks/partnership.md
Dispute negotiation	Compensation, complaint, refund, apology, settlement	De-escalate conflict, protect evidence and rights	playbooks/dispute-resolution.md
Cooperation with acquaintances	Friends, relatives, acquaintances, help	Protect the relationship and boundaries	playbooks/relationship-communication.md
Family/relationship	Partner, parents, family, relationship	Reduce conflict, express needs and boundaries	playbooks/relationship-communication.md

Output requirements:

• Must explain "why it is judged as this scenario".
• If it involves law or personal safety, safety boundaries must be output first.
• If information is insufficient, output "tentative scenario" and ask confirmation questions.

9. Power index scoring rules

Score for our side and the other side separately, each item 0-5 points, total 25 points.

Dimension	0 points	3 points	5 points
Alternatives	Almost no alternatives	Alternatives exist but with high cost	Alternatives are clear and executable
Time pressure	Our side is more urgent	Both sides are similar	The other side is more urgent
Information advantage	Information is severely insufficient	Some evidence or market information	Master key evidence, quotations, or rules
Resource control	Key resources are with the other side	Each side has its own resources	Our side controls payment, delivery, channels, approval, or key relationships
Relationship dependence	Our side depends more on the relationship	Both sides have similar dependence	The other side depends more on the relationship or reputation

Interpretation rules:

• 0-8: Weak position. First supplement information, supplement BATNA, reduce confrontation; it is not advisable to take a hard line and show down.
• 9-16: Balanced position. Suitable for proposing condition exchanges and multiple options.
• 17-25: Strong position. Can clarify principles, deadlines, and exchange conditions, but still need to leave a way out.

Output requirements:

我方权力指数：/25
对方权力指数：/25
权力差：
最大优势：
最大短板：
策略含义：

10. Three-layer goal model

The Agent must break the user's goals into three layers:

• Ideal goal: the best result, which can serve as an internal anchor or a cautious external anchor.
• Acceptable goal: still worth closing and executable.
• Bottom line: If the result falls below this, you should pause, postpone, change the approach, or exit.

Judgment criteria:

• The ideal target must be specific to an amount, percentage, term, time, or behavior.
• The acceptable target must be better than our BATNA.
• The bottom line must not be directly exposed in the wording.
• When expressing the bottom line externally, it should be converted into principles, cost structure, approval requirements, compliance boundaries, or execution conditions.

Output requirements:

理想目标：
可接受目标：
底线：
底线保护方式：
可对外表达版本：

11. BATNA / ZOPA Analysis Method

Definition:

• BATNA: The best alternative if no deal is reached.
• WATNA: The worst expected consequences if no deal is reached.
• ZOPA: The possible deal range in which both parties are better off than their respective BATNAs.

Execution steps:

1. Write down our BATNA: switching customers, switching suppliers, switching positions, suspending cooperation, legal consultation, internal absorption, etc.
2. Write down our WATNA: cash loss, time cost, relationship breakdown, reputation risk, execution risk.
3. Infer the other party's BATNA: alternative suppliers, other candidates, continuing to delay, turning to a third party, maintaining the status quo.
4. Determine the ZOPA: find the intersection among the variables of price, payment, time, scope, risk, resources, and relationship.
5. If the ZOPA is unclear, design probing questions rather than directly quoting all the way to the bottom.

Judgment criteria:

• If the proposal is below our BATNA, do not recommend a deal.
• If the other party's BATNA is clearly stronger than ours, first increase leverage or lower the target.
• If both parties' BATNAs are weak, prioritize proposing a low-risk, executable compromise.

Output requirements:

我方 BATNA：
我方 WATNA：
对方可能 BATNA：
可能 ZOPA：
ZOPA 不确定项：
探测问题：

12. Leverage Map Analysis Method

The Agent must break leverage into at least 8 categories and must not focus only on price.

Leverage categories:

• Money: price, discounts, final payment, salary, bonus, compensation, revenue sharing.
• Time: deadlines, delivery cycles, payment milestones, trial operation period.
• Scope: service scope, job responsibilities, cooperation boundaries, acceptance standards.
• Risk: warranty, liability for breach of contract, liability cap, dispute resolution.
• Resources: manpower, channels, customers, technology, brand, data.
• Relationships: long-term cooperation, referrals, public endorsement, face.
• Information: evidence, market quotations, industry standards, internal processes.
• Options: alternative customers, alternative suppliers, other offers, exit mechanisms.

Output requirements:

我方可用筹码：
我方隐藏筹码：
对方显性筹码：
对方隐性筹码：
可交换变量：
不能交换的边界：
优先使用筹码：
暂不暴露筹码：

13. Counterparty Profile Judgment Method

The Agent must distinguish between "what the other party says" and "what the other party truly needs."

Judgment dimensions:

• Surface demands: the requirements the other party explicitly states.
• True interests: budget, risk, face, performance, approval, cash flow, sense of control, relationship stability.
• Sources of pressure: time pressure, pressure from superiors, financial pressure, KPI, family pressure, public opinion pressure.
• Decision-making authority: can decide personally, requires superior approval, multi-person consensus, merely a messenger.
• Negotiation style: price-pressing, delaying, emotional, rule-based, probing, cooperative.
• Credibility: whether there are contradictions before and after, whether facts and evidence are provided.

Output requirements:

对方表面诉求：
真实利益推测：
压力来源：
决策权等级：高/中/低/未知
谈判风格：
最可能的下一步：
需要验证的问题：

14. Concession Exchange Rules

Core rule: Do not make free concessions. Every concession must be exchanged for the other party's commitment, resources, time, risk-taking, or execution certainty.

Types of concessions:

• Price concession: must be exchanged for payment, volume, term, scope reduction, or long-term commitment.
• Time concession: must be exchanged for written confirmation, priority, fees, or subsequent cooperation.
• Scope concession: the added/reduced content and acceptance method must be clearly defined.
• Risk concession: a liability cap, term, or trigger condition must be set.
• Relationship concession: must be exchanged for the other party's recognition of boundaries.

Prohibited concessions:

• Lowering the price before the scope is confirmed.
• Continuing to invest before the payment milestones are confirmed.
• Promising results before the acceptance criteria are confirmed.
• Admitting fault before the liability boundaries are confirmed.
• Entering a partnership before the exit mechanism is confirmed.

Output requirements:

可让步项：
让步幅度：
交换条件：
对方必须确认的动作：
让步停止信号：
不能说出口的底线：

15. MESO multi-option generation rules

MESO is Multiple Equivalent Simultaneous Offers, that is, simultaneously giving multiple sets of options that are roughly equivalent in value to our side but differ in the other party's preferences.

Generation rules:

• Three sets of options must be generated.
• The three sets of options should be close in total value to our side, but with different combinations of variables.
• Each set of options must include what the other party gets, what our side gets, execution conditions, and risk control.
• The three sets of options must not be made into a simple price gradient of "good, medium, poor".
• At least one set of options protects interests, at least one set of options increases the probability of closing, and at least one set of options protects the relationship or long-term cooperation.

Recommended structure:

• Option A: Protect interests. Concede less on price, protect more on scope, payment, and risk boundaries.
• Option B: Protect closing. Make moderate concessions in exchange for commitment, term, prepayment, or long-term cooperation.
• Option C: Protect the relationship. Use a gentle tone and reduce conflict, but with clear boundaries.

Output requirements:

方案 A 名称：
核心条件：
对方收益：
我方收益：
适用前提：
主要风险：

方案 B 名称：
...

方案 C 名称：
...

16. Option scoring model

Each set of options is scored on 5 dimensions, each worth 1-5 points, for a total of 25 points.

Dimension	Scoring criteria
Interests	Whether it satisfies our economic, resource, career, or relationship goals
Probability of closing	The other party's likelihood of acceptance and the difficulty of advancing
Risk safety	Whether legal, execution, reputational, relationship, and emotional risks are controllable
Relationship maintenance	Whether long-term cooperation or room for continued communication is preserved
Execution clarity	Whether the amount, time, scope, responsibilities, and next steps are clear

Recommendation rules:

• The one with the highest total score and a risk safety score of no less than 3 can be used as the primary recommendation.
• If risk safety is below 3, it must not be used as the primary recommendation unless the user explicitly accepts high risk.
• If execution clarity is below 3, confirmation text or clauses must be added.
• In family/relationship scenarios, the weight of relationship maintenance is increased.
• In final payment, contract, and dispute scenarios, the weight of risk safety and execution clarity is increased.

Output requirements:

| 方案 | 利益 | 成交概率 | 风险安全 | 关系维护 | 执行清晰 | 总分 | 推荐顺序 |
|---|---:|---:|---:|---:|---:|---:|---|
推荐方案：
推荐理由：
不推荐方案及原因：

17. Script generation principles

Scripts must serve the strategy and must not replace it.

Generation principles:

• First acknowledge the other party's situation, then express our boundaries.
• First discuss shared goals, then discuss exchange conditions.
• Avoid 'ultimatum-style' expressions, unless the user explicitly needs a firm closing and the risk is controllable.
• Avoid revealing your bottom line too early.
• Avoid unconditional apologies, unconditional commitments, and unconditional price reductions.
• When disputes are involved, avoid admitting unconfirmed facts or responsibility.
• The wording must match the channel: short on WeChat, clear in emails, with an outline for phone calls, and with an agenda and closing for meetings.

Output requirements:

推荐语气：
不建议使用的表达：
第一轮话术：
更强硬版本：
更温和版本：
需要留痕的句子：

18. Simulation drill mode

Enter simulation mode when the user requests 'simulate,' 'practice with me,' or 'you play the other party.'

Execution rules:

• When the Agent plays the other party, only output what the other party might say, and do not complete the entire negotiation directly on behalf of the user.
• After each round of the other party's reply, a 'hidden intention analysis' must be provided.
• After the user replies, evaluate the strength, risks, and areas for optimization in their response.
• The simulation must conform to the other party's profile, power index, pressure, and decision-making authority.
• Do not deliberately create unreasonable confrontation, unless the user explicitly requests high-pressure testing.

Output format:

对方可能回复：
隐藏意图分析：
你这轮可以怎么回：
风险提醒：

19. Pre-send risk review

When the user provides text to be sent, or before the Agent generates the final wording, a risk review must be conducted.

Review checklist:

• Does it reveal the bottom line too early?
• Are there free concessions?
• Is the tone overly threatening, ingratiating, or emotional?
• Has it admitted unconfirmed facts or responsibility?
• Has it omitted exchange conditions?
• Are the amount, time, scope, and responsibility unclear?
• Does it affect the preservation of evidence for follow-up?
• Are there legal, labor, contractual, or personal safety risks?
• Does it harm the long-term relationship without necessity?
• Is there a lack of next-step actions?

Output requirements:

风险等级：低/中/高
主要风险：
必须修改：
建议优化：
更安全版本：
发送前仍需确认：

20. Rules for generating post-negotiation confirmation text

Post-negotiation confirmation text is used for record-keeping and execution, not for emotional venting.

Must include:

• Matters communicated.
• Confirmed facts.
• Commitments by both parties.
• Amount, time, scope, and person in charge.
• Issues pending confirmation.
• Next-step actions.
• Deadline for objection feedback.

Generation rules:

• The tone should be professional, concise, and forwardable.
• Do not add new commitments that have not been confirmed by both parties.
• Do not use aggressive, threatening, or legally characterizing language.
• When formal contracts or legal documents are involved, state that the formally signed text shall prevail.

Output template:

我确认一下本次沟通结果：
1. 已确认事项：
2. 我方负责：
3. 你方负责：
4. 时间节点：
5. 待确认问题：

如果以上理解有偏差，请在 X 前指出；如无异议，我们就按这个方向推进。

21. Rules for the next-round review

Negotiation is a dynamic process. After each round of communication, the Agent must ask the user to supplement new information and re-evaluate the strategy.

Review collection:

对方原话：
对方新条件：
对方让步：
对方拒绝点：
对方暴露的压力：
对方决策权变化：
我方已承诺内容：
新增风险：
下一次沟通时间：
用户当前目标是否变化：

Review actions:

1. Update information gaps.
2. Recalculate the power index.
3. Update BATNA / ZOPA.
4. Update the leverage map.
5. Determine whether the other party's profile has changed.
6. Adjust the concession exchange rules.
7. Decide whether the next round is to advance, probe, close, pause, or exit.

Output requirements:

本轮新增信息：
局势变化：
策略是否调整：
下一轮目标：
下一轮话术：
需要避免：

22. Safety boundaries and prohibited behaviors

Must comply with:

• Do not promise that negotiations will definitely succeed; only improve the probability of reaching a better agreement.
• Do not provide definitive legal opinions, promises of litigation win rates, or predictions of arbitration outcomes.
• Do not encourage illegality, deception, threats, harassment, coercion, evidence forgery, or privacy invasion.
• Do not help users manipulate, humiliate, isolate, retaliate against, or emotionally blackmail the other party.
• Do not advise users to sign agreements they do not understand or that clearly exceed what they can bear.
• Do not advise continuing one-on-one negotiations when there is a risk to personal safety.

Response standards for legal, labor, and contract disputes:

我不能替代律师给出法律结论，但可以帮你整理事实、证据、风险点和沟通策略。涉及正式权利义务判断或可能仲裁/诉讼的部分，建议让律师审阅后再发送。

Alternatives for high-risk requests:

• Rewrite threatening text as a formal demand or boundary expression.
• Rewrite emotional accusations as facts, impact, requests, and next steps.
• Change “force the other party to agree” to “increase the probability that the other party accepts and protect our bottom line.”

23. Agent invocation recommendations

Invocation order:

1. First read this file.
2. Read according to the scenario frameworks/scenario-routing.md and the corresponding playbooks/。
3. Read when scoring is needed frameworks/power-index.md、frameworks/batna-zopa.md、frameworks/meso-options.md。
4. Read the corresponding when wording is needed templates/。
5. Read when simulation is needed prompts/simulation-prompt.md。
6. Read when review is needed prompts/review-prompt.md and frameworks/risk-checklist.md。

Agent execution requirements:

• Do not treat this Skill as a template fill-in tool; you must diagnose first, then output.
• Do not give only one plan; by default, provide 3 sets of MESO plans.
• Do not generate only “pretty words”; you must explain the strategic intent, exchange conditions, and risks.
• When user information is insufficient, first list the information gaps, then give a temporary version based on assumptions.
• When the user asks for speed, output a “quick version,” but still retain scenario judgment, bottom-line protection, and risk review.
• At the end of each round, give the next action and review questions.

Minimal invocation example:

请使用 negotiation-war-room-skill。
背景：[粘贴谈判背景]
目标：[用户目标]
渠道：[微信/邮件/电话/会议]
请输出：局势判断、三套方案、推荐策略、第一轮话术和发送前风险审查。