bro plugin
Eleven skills to stay aligned with an agent: plain-language restatement, recaps, a status board and a pull back to the goal.
- 179 ★ on the repository
- Last commit
Minimal access
Why this level
- The skills only restate and summarize material; no external services are called.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/backnotprop-broDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skills only restate and summarize material; no external services are called.
This is third-party code. Review the repository files before installing.
What it does
bro is a set of skills that remove misunderstanding between you and an agent during everyday work. `bro` restates the last AI answer in plain language, `readback` rephrases your request to confirm you were understood and `sharpen` clarifies goal, approach and scope before work begins. `clean-room` takes an independent verdict from a fresh agent with no access to the current conversation. For tracking progress there are `facts` with critical facts before or after implementation, `recap` in two or three sentences, `status` with an ASCII board of current work in the terminal and `ladder` with a vertical progress ladder. `big-pics-little-words` and `one-pager` build an HTML page with big visuals and few words for teammates. The latest skill, `converge`, pulls a sprawling session back to the goal, sorts open threads and names the one next thing. Skills install with `npx skills add backnotprop/bro`, and plugins exist for Claude Code and Codex.
Who it is for. Anyone who runs long sessions with an agent and is tired of dense answers, lost threads and restating the task from scratch.
Good fit when
- The agent gave a long technical answer and you want it in plain words.
- Before a big task you want to check the agent understood the goal, approach and limits.
- The session wandered off and you want to return to the main goal and pick the next step.
- An idea must be shown to the team on one HTML page with visuals.
Not a fit when
- You need build or test automation: the skills work on the conversation and not on code.
- Strict code verification is needed: `clean-room` gives an independent opinion and no guarantee.
- You work in one short request with no long session.
Example request
Restate your last answer in plain words with bro, then show the work status as a board with status.Limitations
The skills intervene in the dialog and spend tokens on extra restatements and summaries, so they help little on short tasks. The `clean-room` skill starts a fresh agent with no access to the conversation, so it sees only what you pass it. The `big-pics-little-words` and `one-pager` pages are created as HTML files and must be opened in a browser. Plugins install with separate commands for Claude Code (`/plugin marketplace add backnotprop/bro`) and Codex, and the generic route is `npx skills`. The project is small and documentation is thin.
How to disable. Remove the skills through your manager (`npx skills remove`) or the plugin with `/plugin uninstall bro@bro` in Claude Code.
FAQ
How do I install bro?
With `npx skills add backnotprop/bro`. For Claude Code there is a plugin: `/plugin marketplace add backnotprop/bro`, then `/plugin install bro@bro`.
How do I install the plugin in Codex?
With `codex plugin marketplace add backnotprop/bro` and `codex plugin add bro@bro`.
How does `readback` differ from `sharpen`?
`readback` restates your prompt to check understanding. `sharpen` clarifies goal, approach and scope before work starts.
What does `converge` do?
It pulls a session with tangents back to the goal, names the goal, sorts open threads and picks the one next thing.
What is `clean-room`?
An independent verdict from a fresh agent that cannot see the current conversation and judges only what it was given.
Discussion
No questions or reviews yet.
Related
Plugins/bro
bro
The /bro command re-explains the assistant's last reply in plain language while commands and numbers stay as they were.
A skill that cross-examines a plan before work starts: numbered questions in rounds with a recommended answer and no changes until you confirm
Four skills for a person with several AI assistants: a task register, evidence-based acceptance, beginner instructions
A library of 229 skills for work and life: sales, marketing, finance, engineering, design and parallel agent runs
cafe3310's personal skill collection: spec-driven project work, reviews, reports, browser, 3D and video in one repository
The /myoffice command opens a browser office showing what every agent on your machine is doing: thinking, waiting for you or done.
README in short
The README lists eleven skills with short descriptions, shows `status` output examples and describes installation through `skills`, the Claude Code plugin and the Codex plugin. `converge` is marked as the latest addition.