Deep market research and business analysis plugin for Claude Code and Claude Desktop. Runs under your existing Pro subscription — no API credits required.
Gives Claude 8 specialized tools for business concept research, analysis, and strategy. Claude calls these tools autonomously when you ask it to analyze a concept, suggest ideas, or stress-test a business.
| Tool | What It Does |
|---|---|
run_cbo_analysis |
Full pipeline: research → model → report saved to disk |
run_market_research |
Deep research brief only (12 searches, structured JSON output) |
generate_cbo_report |
Full report from existing brief (skips research) |
suggest_concepts |
3–5 researched concept suggestions based on founder profile |
enhance_concept |
Enhancements, adjacencies, pivots for an existing concept |
voice_of_reason |
Standalone reality check — 6 hard investor questions |
list_research_outputs |
See all saved briefs and reports |
compare_concepts |
Side-by-side scoring of 2–4 concepts |
chmod +x setup.sh
./setup.shThe setup script will:
- Create a Python virtual environment
- Install the
mcpdependency - Print the exact config to add to Claude
claude --plugin-dir /path/to/cbo-researchOr add the MCP server directly:
claude mcp add cbo-research /path/to/cbo-research/.venv/bin/python /path/to/cbo-research/server.pyEdit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"cbo-research": {
"command": "/path/to/cbo-research/.venv/bin/python",
"args": ["/path/to/cbo-research/server.py"]
}
}
}Restart Claude Desktop after saving.
In Claude Code:
claude
> /mcpYou should see cbo-research listed with all 8 tools.
Run a full CBO analysis on [concept]. I'm a serial founder with SaaS and physical
business experience, based in Seattle, ~$200K available capital.
Research the collectibles storage market — I want the brief but not the full report yet.
Suggest 5 business concepts for someone with my background — SaaS, physical businesses,
marketplaces. Interested in collectibles, real estate, and service businesses.
Seattle-based, $150K capital, can run lean.
Run voice of reason on the Stayfront concept. Brief is at output/briefs/[filename].json
Enhance the Collector's Vault concept. Known weakness: slow ramp to breakeven,
single-location revenue ceiling.
Compare Stayfront vs Collector's Vault vs [new concept] side by side.
All outputs save to:
output/
├── briefs/ ← JSON research briefs (timestamped)
│ └── 20260328-143022-collectors-vault.json
└── reports/ ← Full .md reports (timestamped)
└── 20260328-143155-collectors-vault.md
Briefs are reusable — run list_research_outputs before re-researching a concept.
cbo-research/
├── .claude-plugin/
│ └── plugin.json ← Plugin manifest
├── .mcp.json ← MCP server declaration
├── skills/
│ └── market-research-cbo/
│ └── SKILL.md ← Skill definition + trigger rules
├── server.py ← MCP server (entry point)
├── requirements.txt ← Python dependencies (just mcp)
├── setup.sh ← One-time setup script
├── CLAUDE.md ← Claude Code operating instructions
├── README.md ← This file
├── prompts/
│ ├── cbo_system.md ← Master CBO persona + principles
│ ├── research_agent.md ← Research layer instructions
│ └── voice_of_reason.md ← Skeptic/reality-check prompt
└── output/
├── briefs/ ← Auto-created on first run
└── reports/ ← Auto-created on first run
- Edit
prompts/cbo_system.mdto change the CBO persona or core principles - Edit
prompts/research_agent.mdto modify research layers or output format - Edit
prompts/voice_of_reason.mdto adjust the skeptic questions - Add tools to
server.pyby following the existing pattern
This plugin provides both an MCP server (8 tools) and a skill (/market-research-cbo) that triggers automatically when business analysis is discussed. The skill routes to the MCP tools — they work together as a single plugin.