Advisory Board — Claude Code Plugin
Stress-test your decisions with a 6-persona AI advisory board.
A single plugin that takes your decision, question, or proposal and runs it through 6 independent advisors — PM, CMO, CTO, CPO, Operations, and Devil's Advocate. Each evaluates from their distinct lens. You get structured feedback from all six, plus a board synthesis that surfaces the key tension and the single question you need to answer before committing.
What is Claude Code?
Claude Code is Anthropic's free CLI that runs Claude directly in your terminal. It's not a chatbot in a browser tab — it's an AI that reads your files, runs commands, and takes actions. You install it once, and then you can run any plugin from anywhere.
Install Claude Code — takes about 2 minutes.
Install
/plugin marketplace add EagleJD08/advisory-board
/plugin install advisory-board@EagleJD08
Two commands. No Terminal required beyond what Claude Code already has.
Usage
/advisory-board:advisory-board Your decision or question here
First run uses bundled generic board members — no setup needed. Works immediately.
To personalize (recommended after your first run):
/advisory-board:advisory-board setup
This walks you through 5 questions and calibrates the board to your specific company, stage, and goals. A board that knows your context gives sharper, more specific advice.
To update:
/plugin update advisory-board@EagleJD08
The Board
| Seat | Lens | Challenge Style |
|---|
| PM Advisor | Prioritization & sequencing | Asks "compared to what?" on every recommendation |
| CMO Advisor | Audience & positioning | Asks "why would someone care?" relentlessly |
| CTO Advisor | Technical feasibility & leverage | Cuts through hype; strong on build-vs-buy |
| CPO Advisor | Product-market fit | Jobs-to-be-done framing; allergic to assumptions |
| Operations Advisor | Execution risk | Turns every plan into a risk matrix |
| Devil's Advocate | Independent Director | Presents the strongest possible case against your plan |
What It Does
- Parses your decision — identifies decision type, stakes level, and the key assumption you're making
- Runs 6 advisors in parallel — each evaluates independently, with no knowledge of what the others will say
- Produces structured feedback per advisor: verdict, main concern, opportunity, key question, confidence level
- Synthesizes the board — overall sentiment (Aligned / Mixed / Divided), the key tension, the one question you must answer, and the recommended next action
- Learns over time — tracks which decisions you brought to the board and what the outcomes were
See plugins/advisory-board/examples/example-run.md for a full annotated run.
Plugin Structure
advisory-board/
├── .claude-plugin/
│ └── marketplace.json — this repo's plugin registry
├── plugins/advisory-board/
│ ├── .claude-plugin/plugin.json — plugin manifest
│ ├── skills/advisory-board/
│ │ ├── SKILL.md — the skill Claude runs
│ │ └── config.default.md — bundled defaults for first run
│ ├── examples/example-run.md — annotated sample output
│ └── README.md — plugin-level usage docs
├── README.md — this file
└── LICENSE — MIT
The critical structural rule: skills/ lives at the plugin root, not inside .claude-plugin/. This is the most common footgun when building Claude Code plugins.
User config (created on setup, stored locally):
~/.claude/data/advisory-board/config.md
Feedback
Built this because I was making too many big decisions in an echo chamber. The Devil's Advocate alone has saved me from 3 expensive mistakes.
If you try it, I'd love to know what you think. DM me on LinkedIn or open an issue.
License
MIT — see LICENSE.
alirezarezvani
harshilmathur
fatihguner
assafkip
Riley-Coyote
mohitagw15856