feat(growth): restore negotiation maximizer skill

README.md

@@ -1,6 +1,6 @@
 # software-maestro
 
-A capability orchestra of 72 lean agent skills for AI IDE agents (Claude Code, Codex, Antigravity, Gemini, and Hermes Agent).
+A capability orchestra of 74 lean agent skills for AI IDE agents (Claude Code, Codex, Antigravity, Gemini, and Hermes Agent).
 
 ---
 
@@ -36,9 +36,9 @@ A capability orchestra of 72 lean agent skills for AI IDE agents (Claude Code, C
 </details>
 
 <details>
-<summary><strong><code>engineering/</code></strong> (10 skills across 3 subdomains)</summary>
+<summary><strong><code>engineering/</code></strong> (11 skills across 3 subdomains)</summary>
 
-*   **`engineering/development-and-quality/`** (7 skills): Core engineering craft, code review checklists, Test-Driven Development (TDD), evaluation-driven AI development, ASD-STE100 technical writing standards, info-to-ink token compression, and open source contribution etiquette with Conventional Commits (`developer-development-rules`, `developer-code-review-rules`, `developer-test-driven-development`, `developer-eval-driven-development`, `technical-language-rules`, `info-to-ink`, `open-source-contributor`).
+*   **`engineering/development-and-quality/`** (8 skills): Core engineering craft, code review checklists, Test-Driven Development (TDD), evaluation-driven AI development, ASD-STE100 technical writing standards, info-to-ink token compression, open source contribution etiquette, and TypeSafe Jev workflow design (`developer-development-rules`, `developer-code-review-rules`, `developer-test-driven-development`, `developer-eval-driven-development`, `technical-language-rules`, `info-to-ink`, `open-source-contributor`, `typesafe-jev`).
 *   **`engineering/architecture-and-resiliency/`** (2 skills): Clean/Hexagonal system architecture with circuit breakers and zero-key web search grounding (`system-architecture-rules`, `grounded-search-rules`).
 *   **`engineering/cloud-and-infrastructure/`** (1 skill): Multi-cloud architecture, hybrid cloud networking, and reusable Terraform IaC modules (`multi-cloud-architecture`).
 
@@ -92,14 +92,15 @@ A capability orchestra of 72 lean agent skills for AI IDE agents (Claude Code, C
 </details>
 
 <details>
-<summary><strong><code>growth/</code></strong> (6 skills)</summary>
+<summary><strong><code>growth/</code></strong> (7 skills)</summary>
 
 *   **`analytics-event-tracking`**: Behavioral telemetry taxonomies, retention modeling, and CRM lifecycle automation.
 *   **`commerce-ux-rules`**: Shopping cart optimization, checkout friction reduction, and multi-step wizard forms.
 *   **`developer-seo`**: Technical search engine optimization, semantic metadata, and structured schema tags.
 *   **`experimentation-hypothesis-engine`**: Statistical A/B testing, MDE sample sizing, SRM validation, and 3-way post-mortems.
 *   **`conversion-copywriting`**: High-conversion landing page copy, value propositions, and concise prose.
 *   **`american-dialect`**: Research-backed regional American English rewrites that preserve author voice and avoid caricature.
+*   **`negotiation-maximizer`**: Ethical, low-theater negotiation for compensation, purchases, renewals, contracts, and commercial terms.
 
 </details>
 
@@ -143,8 +144,8 @@ A capability orchestra of 72 lean agent skills for AI IDE agents (Claude Code, C
 │   ├── interaction-and-layout/  # Responsive layouts, animation physics, IA wayfinding (3 skills)
 │   ├── experience-and-flows/    # Onboarding UX, Prototype Facade, Chart viz (3 skills)
 │   └── evaluation-and-quality/  # Usability heuristics & 5-Act interview testing (2 skills)
-├── engineering/              # Code Quality, Architecture, Cloud & IaC (10 skills)
-│   ├── development-and-quality/ # TDD, evals, code reviews, craft, STE prose, info-to-ink, OSS etiquette (7 skills)
+├── engineering/              # Code Quality, Architecture, Cloud & IaC (11 skills)
+│   ├── development-and-quality/ # TDD, evals, code reviews, craft, STE prose, info-to-ink, OSS etiquette, Jev (8 skills)
 │   ├── architecture-and-resiliency/ # System architecture, grounded search (2 skills)
 │   └── cloud-and-infrastructure/ # Multi-cloud architecture & Terraform IaC (1 skill)
 ├── finance/                  # Settlement rails, tax nexus, chargebacks & treasury (1 skill)
@@ -153,7 +154,7 @@ A capability orchestra of 72 lean agent skills for AI IDE agents (Claude Code, C
 ├── ops/                      # SEV-0 incident command, crisis comms & blameless RCA (1 skill)
 ├── data-and-api/             # API guidelines, caching, causal inference, semantic layer, metric triage (5 skills)
 ├── security/                 # Developer security & auth/identity rules (2 skills)
-├── growth/                   # Telemetry, commerce UX, SEO, experimentation, regional copy (6 skills)
+├── growth/                   # Telemetry, commerce UX, SEO, experimentation, regional copy, negotiation (7 skills)
 ├── quality/                  # CI/CD, FinOps, dogfooding, a11y, observability, web perf, peekaboo (8 skills)
 └── productivity-maestro/      # Executive memos, meeting intelligence & 3-tier Scrum cadences (5 skills)
     ├── executive-and-async/  # Meeting intelligence & Smart Brevity async memos (2 skills)
@@ -188,7 +189,7 @@ The export copies each skill package, including `references/`, `scripts/`, and `
 
 ## 🔍 Local Skill Discovery CLI
 
-Instantly search, discover, and inspect all 71 local skills from your terminal:
+Instantly search, discover, and inspect all 74 local skills from your terminal:
 
 ```bash
 # Keyword search across names, triggers, and expert frameworks

growth/negotiation-maximizer/SKILL.md

@@ -0,0 +1,156 @@
+---
+name: negotiation-maximizer
+description: Use when the user wants to negotiate a price, rate, compensation, contract, purchase, hotel, service, vendor deal, discount, upgrade, fee, renewal, refund, or other commercial terms; asks what to say to improve leverage or get a better deal; provides an offer/counteroffer and wants a response; or wants to maximize coupons, direct-booking value, bundles, competing configurations, or all-in economics. Do not trigger for ordinary writing, simple price lookup/comparison, arithmetic, generic persuasion, or conflict mediation without a concrete negotiated outcome.
+compatibility: Claude Code, Gemini CLI, OpenAI Codex, and other Agent Skills-compatible runtimes
+---
+
+# Negotiation Maximizer
+
+## Objective
+
+Maximize expected deal value while preserving credibility, optionality, and relationship quality.
+
+Optimize for better economic and non-price terms, lower hidden cost, stronger fallback options, and minimum visible negotiation theater.
+
+**Core rule: maximize leverage; minimize visible tactics.**
+
+The final message should sound like a normal informed buyer or professional making a decision—not someone performing negotiation tactics.
+
+## Default voice
+
+- Casual, concise, direct.
+- Maximize information-to-ink ratio.
+- State exact requirements, dates, quantities, configurations, and decision criteria.
+- Prefer one strong lever over a laundry list of tactics.
+- Keep alternatives credible and usually implicit.
+- Never reveal maximum willingness to pay unless strategically necessary.
+- Make the counterparty move first when that improves information or price discovery.
+- Ask for all-in / including-tax / including-fee pricing when relevant.
+- Signal ability to close without sounding eager, desperate, or theatrical.
+- Avoid corporate-sales phrasing, faux warmth, jargon, and obvious negotiation tricks.
+- If a sentence can be removed without losing leverage or information, cut it.
+
+## Silent preparation
+
+Before drafting, determine:
+
+1. **Objective** — desired outcome.
+2. **BATNA** — best credible alternative.
+3. **Reservation point** — walk-away threshold, if known; keep private.
+4. **Counterparty incentives** — revenue, occupancy, quota, retention, direct-channel economics, certainty, simplicity.
+5. **Information gap** — what should they reveal before the user reveals more?
+6. **Comparison basis** — normalize taxes, fees, configuration, quantity, cancellation, service level, and benefits.
+7. **Leverage** — choose the strongest one or two points only.
+8. **Sequencing** — decide whether to anchor, counter, or make them move first.
+9. **Close path** — give them an easy way to say yes.
+
+Do not expose this analysis unless asked.
+
+## Strategy
+
+### Normalize the real economics
+Compare total value, not headline price: taxes, mandatory fees, included benefits, cancellation, configuration, quantity, payment timing, quality, switching cost, and risk.
+
+### Protect BATNA and private information
+Never invent competing offers. Do not disclose a weak BATNA. Reveal a strong alternative only when it improves the deal. Never anchor with the user's maximum.
+
+### Choose who anchors
+Prefer making the counterparty move first when discount authority is unknown, unpublished pricing may exist, or the user lacks a strong market anchor. Anchor first when objective market data or a credible competing offer gives the user a strong reference point.
+
+### Use one calibrated problem-solving question
+Prefer:
+- “What can you do on the rate to make this the better option?”
+- “What flexibility do you have on the total?”
+- “How close can you get to the alternative?”
+- “How can we structure this so it works at that level?”
+
+One good question beats several stacked questions.
+
+### Trade; do not donate
+Make concessions conditional: “If you can do X, I can book/sign/move forward.”
+
+### Optimize configuration before squeezing price
+Compare one large unit vs. two smaller units, room-only vs. package, monthly vs. annual, price vs. fee waiver/credit/upgrade, direct vs. marketplace, quantity tiers, or removal of low-value add-ons.
+
+### Use seller economics silently
+Infer flexibility from occupancy/capacity, direct-channel commission savings, quotas, retention economics, marginal upgrade cost, perishable inventory, commitment size, and operational simplicity. Use this to choose the ask; do not lecture the seller about it.
+
+### Use loss aversion without theater
+A credible preference is enough: “I’d rather book the two-bedroom if the numbers work.” Never invent deadlines or scarcity.
+
+## Default message architecture
+
+1. Brief context.
+2. Exact desired outcome/configuration.
+3. Strongest comparison or leverage.
+4. One calibrated ask.
+5. Easy path to close.
+
+## Deal-maximization layer
+
+Silently check whether value can come from direct booking, alternate configurations, quantity/length tiers, member or loyalty rates, public promos, targeted offers, card/portal benefits, package removal, fee waivers, credits, upgrades, refundable-vs-nonrefundable spread, price matching, retention offers, or timing.
+
+Do not dump these mechanisms into the message. Pick the strongest lever.
+
+Never recommend false eligibility, fabricated offers, account abuse, obvious pricing-error exploitation, or deception.
+
+## Response modes
+
+- **“What should I say?”** Return the finished message first; keep explanation minimal.
+- **“What’s the strategy?”** Give strongest leverage, information to withhold, first ask, likely counter, and next move.
+- **Offer/counteroffer provided:** normalize total economics, compare to BATNA, recommend accept/counter/wait.
+- **Draft provided:** preserve voice and improve leverage with the fewest edits.
+- **Maximize a deal:** identify alternate structures before assuming price is the only lever.
+
+## Ethics
+
+Never fabricate offers, eligibility, identities, authority, defects, scarcity, or deadlines. Never use coercion, harassment, impersonation, or reputational threats. Strong negotiation is compatible with truthfulness.
+
+## Make [Expert PhD Legend] Proud
+
+Before finalizing, pressure-test the strategy against the canonical thinkers who literally wrote the books practitioners still reference:
+
+- **Roger Fisher, William Ury, Bruce Patton — _Getting to Yes_**: BATNA protected; interests and objective criteria clear.
+- **Howard Raiffa — _The Art and Science of Negotiation_**: alternatives, uncertainty, tradeoffs, and information value handled deliberately.
+- **Max Bazerman — negotiation and behavioral decision research**: anchoring, bias, overconfidence, escalation, and judgment errors checked.
+- **Deepak Malhotra — _Negotiation Genius_ / _Negotiating the Impossible_**: structural leverage, sequencing, hidden interests, and non-price value considered.
+- **Robert Cialdini, PhD — _Influence_ / _Pre-Suasion_**: influence credible, ethical, and subtle—not manipulative.
+- **Neil Rackham — _SPIN Selling_**: the seller-facing logic exposes the real problem and value gap rather than dumping arguments.
+- **Chris Voss — _Never Split the Difference_**: a natural calibrated question or tactical-empathy move lets the counterparty solve the constraint without sounding scripted.
+- **James K. Sebenius — 3-D Negotiation**: setup, sequencing, parties, scope, and deal design improved before bargaining harder.
+
+Final bar:
+1. Fisher/Ury/Patton proud — BATNA protected.
+2. Raiffa/Bazerman proud — economics normalized; biases checked.
+3. Malhotra/Sebenius proud — structure optimized before squeezing price.
+4. Cialdini proud — influence ethical and invisible.
+5. Rackham proud — ask solves a real seller decision problem.
+6. Voss proud — one calibrated question does more work than three arguments.
+7. **User proud** — casual, concise, high information-to-ink ratio, no negotiation cosplay.
+
+If the draft impresses the experts but no longer sounds like the user, rewrite it. **Expert strategy stays invisible; user voice stays visible.**
+
+## Pre-Flight Deal Quality & Leverage Scorecard (DQLS)
+
+Evaluate candidate drafts silently against these gates before returning:
+
+### Hard Safety Gates (Binary: PASS / BLOCK)
+- **Zero Ceiling Leakage**: Never discloses maximum willingness to pay, budget cap, or weak alternatives.
+- **Zero Negotiation Cosplay**: Completely free of negotiation jargon, framework names, and theatrical bargaining speak (no "BATNA", "anchoring", "tactical empathy", or "hostage tactics").
+- **Strict Grounding & Credibility**: No fabricated counteroffers, fake deadlines, false authority, or fictitious competing quotes.
+
+### Strategic Quality Rubric (Target: >= 23 / 25)
+
+| Dimension | 1 - Critical Failure | 3 - Competent / Generic | 5 - Expert / Invisible Leverage |
+| :--- | :--- | :--- | :--- |
+| **Normalized Economics** | Fixates solely on headline rate. | Mentions taxes/fees but compares apples to oranges. | Compares true all-in total; tests alternate configurations (tiers, units, credits, direct-booking margin). |
+| **Calibrated Burden** | Demands a flat discount or makes aggressive demands. | Asks multiple questions or vague favors. | Uses **exactly one** calibrated question that forces the counterparty to solve the pricing constraint. |
+| **Conditional Trade** | Donates concessions for free or begs for price relief. | Mentions moving forward loosely without a firm condition. | Clean conditional trade: *"If you can do X, I can book/sign/close today."* |
+| **Information-to-Ink** | Wordy, over-explains context, uses faux warmth/filler. | Standard professional email tone. | Maximum density: every word carries leverage; removes anything that doesn't advance the close. |
+| **Frictionless Close** | Open-ended dead end with no clear next step. | Asks for a callback or generic follow-up. | One-click close path: gives the seller the simplest possible way to say "yes". |
+
+### Execution Gate
+- **Score < 20 or any Hard Gate FAIL**: Rewrite immediately.
+- **Score >= 23**: Deliver draft (message first; keep scorecard and strategy notes private unless requested).
+
+Read `references/source.md` when deeper rationale or attribution is needed.

growth/negotiation-maximizer/agents/openai.yaml

@@ -0,0 +1,7 @@
+interface:
+  display_name: "Negotiation Maximizer"
+  short_description: "Negotiate smarter while keeping tactics invisible"
+default_prompt: "Use $negotiation-maximizer to improve this negotiation, identify leverage, and draft a concise message in my voice."
+
+policy:
+  allow_implicit_invocation: true

growth/negotiation-maximizer/evals/cases.json

@@ -0,0 +1,45 @@
+{
+  "skill": "negotiation-maximizer",
+  "version": 1,
+  "trigger_cases": [
+    {"prompt": "Help me get this hotel to discount the two-bedroom suite.", "trigger": true},
+    {"prompt": "Write a recruiter response; I want more comp without giving my floor.", "trigger": true},
+    {"prompt": "They offered 15% off annual SaaS. What should I counter?", "trigger": true},
+    {"prompt": "The 2BR costs more than two 1BRs. Help me use that.", "trigger": true},
+    {"prompt": "Rewrite this so they name the discount first.", "trigger": true},
+    {"prompt": "Convert 62,935 MXN to USD.", "trigger": false},
+    {"prompt": "Summarize this hotel review.", "trigger": false},
+    {"prompt": "Find current coupon codes for this store.", "trigger": false},
+    {"prompt": "Rewrite this thank-you email.", "trigger": false},
+    {"prompt": "What is the cancellation policy?", "trigger": false}
+  ],
+  "binary_cases": [
+    {
+      "id": "protect-ceiling",
+      "prompt": "Draft a hotel negotiation email. My maximum is $650/night but I want them to quote first.",
+      "pass_rule": "The message asks the hotel to move first and does not disclose the user's $650 ceiling.",
+      "must": [["what can you do", "best rate", "flexibility"]],
+      "must_not": ["$650", "650/night", "my maximum"],
+      "pass_fixture": "What can you do on the all-in two-bedroom rate to make it the better option for the stay?",
+      "fail_fixture": "My maximum is $650 per night. Can you meet it?"
+    },
+    {
+      "id": "all-in-economics",
+      "prompt": "Help me compare a direct hotel offer against another configuration with fees and taxes.",
+      "pass_rule": "The response normalizes the comparison to total/all-in economics before recommending a move.",
+      "must": [["all-in", "including tax", "total"], ["compare", "comparison"]],
+      "must_not": ["headline rate only"],
+      "pass_fixture": "Compare the two options on an all-in basis including tax and mandatory fees, then use the cheaper equivalent configuration as leverage.",
+      "fail_fixture": "Compare only the advertised nightly rates and ignore taxes and fees."
+    },
+    {
+      "id": "invisible-tactics",
+      "prompt": "Make this negotiation email sound like me but stronger.",
+      "pass_rule": "The draft is concise and natural, uses at most one primary calibrated ask, and does not name negotiation frameworks or tactics.",
+      "must": [["what can you do", "what flexibility", "how close"]],
+      "must_not": ["BATNA", "anchoring strategy", "tactical empathy", "hostage negotiation"],
+      "pass_fixture": "What can you do on the two-bedroom rate, including tax, to make it the better value for the stay? If the numbers work, I'd prefer the two-bedroom and can book direct.",
+      "fail_fixture": "Using my BATNA and a calibrated-question hostage-negotiation tactic, I want to anchor you lower."
+    }
+  ]
+}