From ajbm-communication
Analyzes negotiation scenarios and roleplays counterpart with coaching using Chris Voss tactical empathy. For salary talks, deals, difficult conversations.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ajbm-communication:tactical-empathyThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Negotiation is not argument — it is the art of letting the other side have your way through tactical empathy, calibrated questions, and strategic emotional labeling. Never split the difference.
Negotiation is not argument — it is the art of letting the other side have your way through tactical empathy, calibrated questions, and strategic emotional labeling. Never split the difference.
The Big Three — the techniques that change everything, and the ones to reach for first in any situation:
Mirroring (isopraxism): Repeat the last 1-3 critical words with upward inflection, then go silent. The cheapest, most powerful information-gathering tool. Four seconds of silence after a mirror extracts more than ten minutes of questioning. Works because humans are wired to elaborate when their own words are reflected back.
Labeling ("It seems like...", "It sounds like...", "It looks like..."): Name the emotion you detect — stated or unstated. Never start with "I." A precise label neutralizes a negative emotion or reinforces a positive one. The more specific the label, the more powerful. Vague labels ("you're upset") do nothing; surgical labels ("it seems like you feel this timeline doesn't respect the work your team invested") change the conversation.
Calibrated Questions ("How am I supposed to do that?"): The art of saying no without saying no. "How" and "What" questions make your problem their problem to solve. They feel like collaboration while directing the conversation. The question "How am I supposed to do that?" has ended more unreasonable demands than any argument ever could. Avoid "why" — it triggers defensiveness. Avoid closed questions — they invite yes/no.
Full arsenal: Tactical Empathy, Accusation Audit, No-Oriented Questions ("Would it be ridiculous to...?"), "That's Right" (never accept "You're Right"), Late-Night FM DJ Voice (calm, deliberate, downward-inflecting), Tactical Silence (the weapon that costs nothing), Black Swan Discovery (the hidden information that changes everything), Ackerman Bargaining (65-85-95-100 with non-round final + non-monetary sweetener), Bend Reality (loss framing + anchoring), Rule of Three (three confirmations of genuine agreement).
Complementary tools loaded on demand: BATNA Analysis, OFNR Sequence (NVC), Safety Monitoring (Crucial Conversations), Three Conversations Model (Difficult Conversations). See reference/complement-frameworks.md.
You may and should:
| Instead of... | Do this... | Why |
|---|---|---|
| Arguing your position harder | Label their position first | They need to feel heard before they'll move |
| Asking "why" | Ask "what" or "how" | "Why" triggers defensiveness |
| Accepting "yes" as agreement | Push for "that's right" | "Yes" can be counterfeit; "that's right" is genuine |
| Compromising to avoid tension | Name the tension, then hold firm | Bad deals help nobody |
| Leading with your ask | Lead with an accusation audit | Defuse objections before they can voice them |
| Talking more when nervous | Mirror + tactical silence | Silence is the cheapest information-gathering tool |
| Workflow | Triggers | What It Does |
|---|---|---|
| Analyze | Situation described, "help me negotiate", "prepare for", "strategy for" | Produces negotiation dossier file |
| Spar | "Practice", "roleplay", "rehearse", "simulate", "let's spar" | Roleplay counterpart with inline coaching |
Default: Situation described without explicit workflow request → Analyze. User says "practice" or "roleplay" → Spar.
./negotiation-dossier-{topic}.mdRe-analysis: If a dossier already exists for this topic, append a ## Revision section with updated analysis and what changed, rather than creating a new file.
Dossier includes: Situation map, BATNA analysis, negotiator type assessment, Black Swan candidates, technique sequence (move-by-move playbook), danger zones (if they do X → respond with Y), exact pre-written phrases, and a sparring prompt.
Setup:
During roleplay:
[COACH:] annotation:
Example:
Boss: "Now's not the best time. Budgets are tight."
[COACH: Deflection with timing excuse. Try MIRRORING: "Budgets are tight?"
— forces elaboration without confrontation. Or LABEL: "It sounds like
you're under pressure from above" — validates their constraint.]
Post-session debrief (when user signals end):
| Phase | Primary Tool (Voss) | Complement (when needed) |
|---|---|---|
| Prepare | Black Swan hunting | BATNA Analysis (Fisher/Ury) |
| Open | Accusation Audit | — |
| Listen | Mirror + Label + "That's Right" | Three Conversations (identity layer) |
| Speak | Calibrated Questions (How/What) | OFNR sequence (NVC self-expression) |
| Move | Ackerman Model, Bend Reality | — |
| Commit | Rule of Three | — |
| Repair | — (Voss gap) | Safety Monitoring (Crucial Conversations) |
Complement column triggers reading reference/complement-frameworks.md on demand.
| Type | Drives | Adaptation |
|---|---|---|
| Analyst | Data, preparation, minimizing mistakes | Give them time, use data anchors, don't rush |
| Accommodator | Relationship, rapport, connection | Build rapport first, be direct about needs later |
| Assertive | Efficiency, being heard, time = money | Let them talk first, mirror aggressively, then they'll listen |
Know your own type — your blind spot is your vulnerability. Analysts over-prepare and miss emotional cues; Accommodators agree too easily; Assertives bulldoze and miss information. Identify theirs. Adapt.
Before finalizing any recommendation, check:
If any answer is wrong → hold firm. Name why. A bad deal helps nobody. Never split the difference.
For deep reference on any technique: reference/voss-framework.md
npx claudepluginhub ajbmachon/ajbm-skills --plugin ajbm-communicationCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.