From vallor
Walks a category manager or sourcing manager through the classic 7-step negotiation process for a specific deal — Prepare, Set Targets, Build the Concession Plan, Open, Bargain, Close, Measure. Each step produces an artifact (target sheet, opening position, concession log, post-mortem). Persists state in the matter / category folder so a paused negotiation can resume mid-step. Use when the user says "negotiation plan", "walk me through the negotiation", "7-step", "structured negotiation", "I'm negotiating with X, prep me", or wants a step-by-step process rather than a single research packet (which is /negotiation-report).
How this skill is triggered — by the user, by Claude, or both
Slash command
/vallor:7-step-negotiation <counterparty> [--step 1-7] [--category <category>]<counterparty> [--step 1-7] [--category <category>]The summary Claude sees in its skill listing — used to decide when to auto-load this skill
A negotiation goes well when the prep is done before the first meeting. This
A negotiation goes well when the prep is done before the first meeting. This skill structures the prep, the bargaining, and the post-mortem into seven discrete steps. Each step produces an artifact. State persists — pick up where you left off.
~/.claude/plugins/config/vallor/CLAUDE.md →
## Negotiation posture. Apply the user's default approach, concession
discipline, walk-away authority.--category is set, load that category's strategy from
category-strategies/ if it exists.Use the AskUserQuestion tool to gather optional context that materially changes the negotiation plan. The goal is a tailored, higher-quality output, not interrogation.
Up front (high-level, before heavy research): Ask 1-3 short questions about anything you cannot infer from the inputs, the practice profile, or the category strategy. Examples for this skill:
Mid-flight (after doing some work): After the Step 1 knowledge pack, after market research, or once you see the counterparty's pressure points, you will surface concrete decisions the user has not settled. Stop and ask before guessing on:
Rules:
Per negotiation, write to:
~/.claude/plugins/config/vallor/negotiations/<YYYY-MM-DD>-<counterparty>-<category>/state.yaml
The state file tracks:
On invocation, the skill reads the state file (if it exists) and resumes at the
current step. Pass --step N to jump to a specific step.
Goal: know more than the other side.
Build a knowledge pack:
Artifact: prepare.md — a structured knowledge pack.
Use web search for external research. Tag every claim with source.
Output of Step 1. Knowledge pack saved. Decision tree:
- Move to Step 2 — set targets.
- Build the full negotiation packet —
/vallor:negotiation-reportproduces a richer artifact with charts and email drafts.- Stop here — knowledge pack is reusable across negotiations.
Goal: know what success looks like before you walk into the room.
For each issue:
| Issue | Aspirational (open here) | Target (would close at) | Walk-away (won't sign below) |
|---|
Cover at minimum:
Each target is justified — "Aspirational price is 12% below current because the benchmark shows competitors at 8-15% lower for similar volume." Aspirational without justification is a wish, not a target.
Walk-away authority check. If any walk-away exceeds the user's authority per the practice profile's escalation matrix, surface it:
"Your walk-away on price ($X reduction) is within your authority ($Y threshold), but the walk-away on liability is not — that needs GC sign-off before we open. Want me to draft the escalation note?"
Artifact: targets.md — the target sheet.
Goal: know what you'll trade before you sit down.
For each issue, define:
A good concession plan answers: "If they push hardest on price, what's the chip I trade?"
Artifact: concession-plan.md — a matrix of moves and reciprocal asks.
Goal: open at the aspirational position, not the target.
Drafts the opening communication — usually an email or term sheet. Tone is set
from the practice profile's Default approach.
Artifact: opening.md — the opening position document or email draft.
After the user sends and gets a response, log the response in
session-log.md and move to Step 5.
Goal: structured back-and-forth, tracked in a concession log.
For each round:
Artifact: concession-log.md — chronological log of every move.
A round-end check: "We've now conceded $X total. That's [Y%] of the aspirational delta. We're still [above/at/below] target. Continue or hold?"
When the walk-away line is crossed: stop the user. "We're about to cross the walk-away on [issue]. This is a stop-and-think moment. Three options: (a) hold the line and risk no deal; (b) escalate to [approver from matrix] for a new walk-away; (c) signal we're at our final position and offer one closing sweetener."
Goal: lock the agreement before it drifts.
Produces a deal memo summarizing every agreed term, in plain English, for both sides to sign off on BEFORE the long-form contract is drafted. This is the "this is what we shook on" document that prevents drafting drift.
Artifact: deal-memo.md — every agreed term, both sides sign or initial.
Handoff to drafting: /vallor:contract-prep --brief <Contract Brief> or directly to /vallor:contract-redline-docx if the
counterparty is on their paper.
Goal: post-mortem and feed forward.
Two weeks after signing, capture:
Artifact: post-mortem.md — feeds back into category strategy + practice
profile (the playbook-monitor agent picks up patterns over time).
The post-mortem is the single highest-leverage step. Skipping it means the same mistake gets made next quarter.
/vallor:negotiation-report produces the parallel
research-heavy artifact for the same deal./vallor:7-step-negotiation Acme
/vallor:7-step-negotiation Megacorp --step 5
/vallor:7-step-negotiation Acme --category "IT Hardware Maintenance"
npx claudepluginhub boardai/vallor-plugins --plugin vallorGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.