Issue #5: Try this this week: give a brand its context

Fifth issue, same shape: one skill to try, what we shipped, one question for you.

Try this this week: give a brand its context

The problem. Everything you know about your brands lives inside a chat window. You explain the account to Claude, it gets sharp for an hour, the session closes, and the knowledge is gone. You type it again tomorrow. You type it again if you move to a different model. Every account manager on your team types their own version, because none of them can see yours. What you know about a brand should not be trapped in one person's chat history with one AI.

The solution. A brand context writes that knowledge down once, outside the chat, where any skill and any teammate can read it: seller and vendor accounts, ACOS targets, item groups, brand terms, negation rules, the structural events behind the strange months, and how you want the brand reported. mx-brand-context fills in what your trailing history can answer, asks you for the rest, and hands you a page to correct before anything downstream uses it. It publishes to your org store as it writes, so the brand belongs to your team rather than to one session.

Try this: open Claude with the MixShift plugin and say:

"set up brand context for <brand>"

Answer the questions, then fix what is wrong on the page it gives you. Every health check, negation pass, and monthly report after that reads the brand the way you do.

Activate this skill →

This week in MixShift

What we shipped Friday in 0.8.10. Full notes at mixshift.ai/releases; deeper how-to lives in the knowledge base.

1. Hand an account to a teammate without setting it up twice

You build the brand context, a teammate picks the account up on their own machine, and they start from nothing. Now the context arrives on its own the first time any skill touches an account your team set up elsewhere. Signing in also tells you how many brands your org has configured and how many are missing here.

Now you can: move an account between people mid-week without a setup command; it just works.

2. Sub-brand discovery is fixed; sorts by largest first

You open a promotion plan to decide which brands are worth their own setup. It has to be complete, and the money has to be at the top. Neither was true. If one of the reads behind the plan failed, the plan was built anyway, so "no campaigns yet" could look like a finding when it was really a missing answer. It now stops and says a read did not complete, and brief drops retry on their own. Ordering was by item count, so a big catalog of dead SKUs could outrank the label carrying most of the revenue. It is now trailing revenue plus ad spend, largest first.

Now you can: read the plan top down and trust that the top is where the money is.

The current plugin version is 0.8.10. To check whether you are behind, open the plugin and say "what's new with MixShift?"

What can we solve for you?

Reply to this email. It goes to MixShift, and every reply gets a personal answer. If we already solve your problem, we point you at the skill. If we build it, we ship it with your name on the release note.

---

Don't have the plugin yet? Get it here.

Talk next week.

MixShift