⏳ This skill is pending AI review.
Scores will appear once the review pipeline completes.
negotiation-playbook
>
Choose how to use this skill
You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.
1. Native installer
This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.
Do not guess an installer command or replace an existing version without reviewing the diff.
2. Complete package recommended
Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.
No complete ProSkills package is published for this listing yet.3. Prompt-only
Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.
Need only the instruction file?
Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.
No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.
// RATINGS
Not yet listed on ClawHub or SkillsMP
// README
Creator Management Inbox
Your creator inbox, drafted in your voice. Claude reads your unread partnership emails, decides what an experienced influencer marketing manager would do, and leaves a ready-to-send draft under each one in Gmail. Rate negotiations, script feedback, go-live instructions, renewals, reschedules, declines. It never sends anything. You glance, edit if you want, and hit send.
Built for people who run creator partnerships - brand-side partnership managers, founders doing influencer marketing themselves, and anyone answering creators, UGC makers and affiliates who is short on time.
Try it in about 15 minutes without connecting anything. The demo answers six fictional creator emails (a rate counter, a script review, a renewal, a reschedule, an affiliate question and an email that tries to trick the AI), so you can judge the drafts before you give it access to anything. Sell physical products? Ask Claude to switch to the e-commerce demo for eight emails to a skincare brand: seeding, a late post, usage rights, an unpaid invoice, a payment scam.
What you get
| Inbox auto-draft | Reads every unread email, sorts it (admin -> content -> money), drafts the reply you'd have written, threads it in Gmail, leaves the original unread, and gives you a one-page report |
| Negotiation playbook | Adapts to how your program runs - always-on or campaigns, per-creator fees or a budget pot, sales or reach or UGC - on top of a full rates-and-terms doctrine: the two-piece first deal, never a bare number, Voss-style calibrated questions, the re-scope counter, renewals, recalibrations, graceful exits |
| Rate engine | OPEN / TARGET / private MAX for every negotiation, from your rate bands - set up by interview or built from your past deals |
| Your voice | Reads your sent emails and interviews you for 5 minutes, then every draft sounds like you. House rules (no em dashes, words you never use) enforced automatically |
| Content reviewer | Reviews scripts and videos against a performance rubric and writes the feedback in your voice |
| Creator notes | One private note per creator: history, results, preferences, fees - so replies and renewals start from facts, not memory |
| Safety built in | Drafts only. Emails are treated as untrusted. A guard blocks send tools. Details in SECURITY.md |
It gets better the more you use it: each run compares its drafts with what you actually sent and proposes rules from your edits. Target: 90%+ of drafts sent without a single change.
How it works (30 seconds)
flowchart LR
A[Unread emails] --> B{Sort by stakes}
B -->|admin, thanks, dates| C[Block 1: light replies]
B -->|script or video| D[Block 2: content review]
B -->|rates, renewals, terms| E[Block 3: negotiation + rate engine]
B -->|legal, disputes, suspicious| F[Block 0: flagged for you]
C & D & E --> G[Safety checks]
G --> H[Draft in Gmail, threaded, original left unread]
H --> I[You review and send]
Full walkthrough: docs/HOW-IT-WORKS.md.
Get started
What you need
- Claude Code - in the Claude desktop app (Code tab) or the terminal. Install guide.
- A Mac, Linux, or Windows with WSL, and Python 3.9+. A Mac has it, but the first use may ask to install Apple's free developer tools: click Install and wait a few minutes. Claude will help on other systems.
- Gmail (Google Workspace or personal). Outlook is planned for v1.1; Your Voice can already learn from an Outlook connector.
- About 15 minutes for the demo, 30-45 minutes for the full setup.
Step 1 - Install (2 minutes)
In Claude Code, type these two commands:
/plugin marketplace add hakon-junge/creator-management-inbox-skills
/plugin install creator-management-inbox@creator-management-inbox-skills
Then turn on updates so you get improvements automatically: type /plugin,
open Marketplaces, select creator-management-inbox-skills, choose Enable
auto-update. Run /reload-plugins (or restart Claude Code).
Step 2 - Hand it to Claude
Paste this into Claude Code:
Set up my Creator Management Inbox. Start with the demo so I can see the drafts, then
walk me through connecting Gmail, my profile, my voice and my rates, one step at
a time. I'm not very technical, so explain each step in plain words.
That's it. Claude runs the inbox-setup skill and guides you through:
| Stage | Time | What happens |
|---|---|---|
| 1. Demo | 15 min | Six fictional emails answered (eight in the e-commerce demo); open the drafts in your browser |
| 2. Connect Gmail | 15 min | Your own private Google connection (guide) |
| 3. Program + profile | 15-25 min | How your program runs (always-on or campaigns, budget, what success means, how you set fees), then your company facts from your website |
| 4. Your voice | 10 min | Built from your sent emails + a short interview |
| 5. Your rates | 10 min | From your past deals, or a guided interview |
| 6. First real run | 5 min | Drafts for your 5 newest unread emails, in Gmail |
Step 3 - Use it
| Say | What happens |
|---|---|
run my inbox | Drafts replies to every unread email, then reports |
run my inbox, only the 5 newest | Same, limited |
reply to the email from Maya | One thread |
what should I offer Maya? 41k views, US audience, she asked for $2,400 | A rate card and the recommended move |
review this script + paste | Approve / request edits, feedback in your voice |
add Leo to my creator notes | Creates his note from your email history |
build my voice / set up my rates | Improve drafts / numbers any time |
is everything set up? | Health check with the fix for anything missing |
You are the sender
Every draft is a suggestion. You read it, you decide, you press send, and what you send is yours: the numbers, the terms, the promises. Check every money email before it goes out. The tool is provided as-is under the MIT licence, with no warranty and no guarantee that a draft is correct.
Support
Maintained by a small team, best effort. Supported today: Gmail on macOS, in Claude Code. Linux and Windows (WSL) should work but are less tested; Outlook is planned for v1.1. Bugs and setup problems go in Issues, questions in Discussions. We read everything and prioritise what affects the most people; we can't offer one-to-one setup help or response times.
Your data stays yours
Everything personal lives in one private folder on your computer,
~/.claude/inbox/, readable only by your user account. It is never inside the
plugin, so updates never touch it, and it's never uploaded anywhere by this tool.
~/.claude/inbox/
profile/ your program, company facts, voice, rates, settings (plain text, edit freely)
creators/ one note per creator
attachments/ the only folder drafts may attach files from
runs/ what was drafted, for learning (text auto-deleted after 14 days)
token.json your Gmail connection (never read by Claude directly)
Please know: to write a reply, Claude reads the email. That means email content is processed by Anthropic's Claude models under your Claude account's terms. If your company has an AI policy, check it first. See SECURITY.md.
Levels
- Level 1 - Out of the box: demo, Gmail, your program and profile. Good admin and content replies; money emails ask creators for their stats instead of naming fees.
- Level 2 - Sounds like you: your voice profile. The biggest single jump in draft quality.
- Level 3 - Negotiates for you: your rate bands and creator notes. Money emails name the right number with the right mechanism.
- **Level 4 - Y
// HOW IT'S BUILT
KEY FILES