Free download
Get the Backbrief Kit, free.
The Backbrief Kit is a free multi-agent framework for Claude Code: an orchestrator that plans the work and routes each piece to a specialist team, and a fresh-context verifier that checks every result before it ships. Plain Markdown, nothing to build, nothing to run. Your project starts running like a team, not one endless chat.
No prompting skill required. /setup interviews you, plain language states the goal, and the orchestrator writes the plan. If you are ever unsure, type /next.
Page updated August 22, 2026. The changelog in the download names the current release.
What you get
An orchestrator that reads your goal and context, writes the plan to a file, and routes the work.
Specialist agents that each own one domain, plus a cheap runner that takes mechanical work at lower cost.
The
/critiqueand/councilcommands for stress-testing decisions from several angles in one pass.The
/grade-ideacommand, which scores a raw business idea on six dimensions against a public 8.5 pass line and names your weakest dimensions.The
/nextcommand, so you never wonder what to do next: it reads your project files, tells you where you stand, and offers the next step. Correct even in a brand-new chat, because it reads what is on disk, not what either of you remembers.The rules that keep the team routing, verifying, and shipping instead of drifting back into one long chat.
A knowledgebase for your own documents and links: the researcher cites your sources ahead of general knowledge, and naming a document in chat is enough to have it filed.
A scaffold of working folders you copy into each project, so state lives in files instead of your memory.
Fresh start or already building?
Either way, the kit installs without touching what you have. If you are starting clean, copy it in and go. If you already run your own agents, rules, or a CLAUDE.md, the included INTEGRATION.md shows how to add the kit alongside your setup without overwriting anything.
Came from The 8.5 Challenge?
You are in the right place. /grade-idea ships in this kit, free. Install it, type /grade-idea, describe your idea in a sentence or two, and you get a scorecard: a score per dimension, the average, the 8.5 pass line, a letter grade, and your lowest dimension named plainly. When several dimensions tie at the bottom, it names all of them.
The pass line is a stress-test, not a prediction. A low score is the useful result.
Post your score with #8point5challenge. The number is the hook, but your lowest dimension is the useful part, so post that too.
Thin descriptions cap what can be scored. When that happens the scorecard says so and names the gap, rather than inventing six confident numbers to fill the space.
Get the free kit
Enter your email and we send the download link. No spam. Unsubscribe anytime.
The install is one instruction.
Claude installs it for you. Open Claude in the unzipped folder, type install this, and it asks where your project lives, copies the files, runs the checks, and never touches anything you wrote.
Fastest install: the Claude Code plugin
Same kit, no download. Run these three commands in Claude Code and the setup interview takes it from there.
/plugin marketplace add charlesashe/backbrief-kit/plugin install backbrief-kit@backbrief/backbrief-kit:setup
Prefer the zip? The form above emails you the same kit as a download, and the one-instruction install applies.
Common questions
What is the Backbrief Kit?
The Backbrief Kit is a free multi-agent framework for Claude Code. It ships an agent team (an orchestrator, specialists, and an independent verifier), commands including /grade-idea, always-on rules, and bundled skills as plain Markdown files. It makes Claude Code run a project like a coordinated team instead of one long chat.
How do I set up a team of agents in Claude Code?
Install the kit and the team is set up for you: unzip the download, open Claude Code in the folder, and type install this. Claude copies the files, runs the checks, and never touches anything you wrote. Or add it as a plugin: /plugin marketplace add charlesashe/backbrief-kit, then /plugin install backbrief-kit@backbrief. Either path ends at /setup, a short interview.
Do I need to know how to prompt?
No. /setup interviews you, plain language states the goal, and the orchestrator writes the plan. If you are ever unsure what to do, type /next: it reads your project files and offers the next step, even in a brand-new chat.
How is this different from writing my own subagents?
Rolling your own gives you full control, and for engineers who want custom behavior it is the right call. The kit encodes the parts most home-grown setups skip: routing rules, cost discipline across models, state that lives in files instead of chats, and a verifier that reads finished work in fresh context.
Will it overwrite my existing Claude Code setup?
No. The install is additive: it never touches your files, and the included INTEGRATION.md shows how the kit sits alongside existing agents, rules, or a CLAUDE.md you already run.
What is the difference between the kit and Backbrief Business OS?
The kit is free and complete on its own, and it includes /grade-idea: the five-advisor panel scoring a raw idea on six dimensions. Backbrief Business OS ($79 one-time) includes the full kit and adds the business layer on top: a CEO loop that turns a brief into a plan with 90-day execution units and grades that plan against the same public rubric with a capped revise loop, plus an operations tier that stays locked until you record a go. The free kit also ships the escalation rule as an instruction your team follows; Backbrief Business OS ships the same rule plus an optional permission layer, offered at install and never added to a setup you already run without your yes, that stops matching outward shell commands at Claude Code's own permission engine.
The kit is the base. Here is what builds on it.
The kit is free and complete on its own, and that includes /grade-idea: five advisors, six dimensions, one number for a raw idea. When the project you are running is a real business, Backbrief Business OS adds the business layer on top of it: the CEO loop that turns a brief into a plan with 90-day execution units, then grades that plan against the same public rubric with a capped revise loop that regrades after you fix the weakest dimensions. The grade is a stress-test, not a success promise, in both products. Everything we announce, we announce to kit owners first.