Brings Clayton Christensen's strategic frameworks into Claude through three tools: analyze_decision runs full disruption theory analysis on business choices, case_study matches your situation to canonical examples like steel minimills or disk drives, and get_framework surfaces jobs-to-be-done or capabilities-processes-priorities theory on demand. The server structures prompts to maintain Christensen's voice and analytical approach, asking about customer jobs, asymmetric motivation, and organizational constraints rather than giving generic advice. Reach for this when you need theory-grounded strategic thinking on market entry, product positioning, or capability decisions. Built as a proof-of-concept for persona agents with validation tests to keep outputs faithful to Christensen's actual frameworks and case studies.
A Clayton Christensen persona agent for strategic advisory, implemented as an MCP (Model Context Protocol) server.
Apply disruption theory, jobs-to-be-done, and capabilities analysis to your business decisions.
This MCP server gives Claude the ability to analyze decisions through Clayton Christensen's frameworks:
When you connect this server to Claude Desktop or Claude Code, you get a strategic advisor that thinks like Christensen—asking the right questions, connecting to historical patterns, and offering theory-grounded insights with appropriate humility.
# Clone the repository
git clone https://github.com/m2ai-mcp-servers/mcp-christensen.git
cd mcp-christensen
# Install dependencies
npm install
# Build
npm run build
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on Mac or %APPDATA%\Claude\claude_desktop_config.json on Windows):
{
"mcpServers": {
"christensen": {
"command": "node",
"args": ["/path/to/christensen-mcp/dist/index.js"]
}
}
}
Add to your project's .mcp.json:
{
"mcpServers": {
"christensen": {
"command": "node",
"args": ["./path/to/christensen-mcp/dist/index.js"]
}
}
}
analyze_decisionFull strategic analysis using all Christensen frameworks.
Use the analyze_decision tool to evaluate: Should we pursue enterprise customers or focus on SMBs?
Parameters:
decision (required): The business decision to analyzecontext: Additional context about your situationorganization: Your org type, size, industry, current focusmarket: Incumbents, competitors, customer segmentsfocusAreas: Specific frameworks to apply (jtbd, disruption, cpp, resource-dependence)case_studyExplore Christensen's canonical case studies and match them to your situation.
Use the case_study tool to find patterns matching: We're a low-cost entrant being ignored by incumbents
Available cases:
steel_minimills: Low-end disruption with asymmetric motivationdisk_drives: New-market disruption with different value networkmilkshake: Jobs-to-be-done discovery through circumstancehonda_motorcycles: Emergent strategy through market learningintel_microprocessors: Capability migration through process investmentget_frameworkLearn about any Christensen framework in detail.
Use the get_framework tool with framework: jtbd and detail: full
Frameworks: jtbd, disruption, cpp, resource-dependence, all
Detail levels: summary, full, questions-only
You: I'm trying to decide whether to build individual AI agents or invest in a factory process for creating them. Can you analyze this using Christensen's frameworks?
Claude: Uses analyze_decision tool
Let me think through this using Christensen's frameworks...
Jobs-to-Be-Done: What job are your customers hiring an AI agent to do? If they need specialized expertise (functional), confidence in the advice (emotional), and credibility (social), individual high-quality agents might serve that job better. But if they need to create their own agents quickly, the factory serves a different job entirely.
The Intel Pattern: I'm reminded of Intel's pivot from memory to microprocessors. The insight wasn't about resources—Japanese firms could acquire those. It was about processes. Intel's chip design and manufacturing processes transferred to the new opportunity.
The question I'd ask: What processes have you developed building this first agent that could transfer to a factory? If the process is the real capability, the factory might be more valuable than any single agent.
You: We're entering a market with a simpler, cheaper product. Incumbents seem happy to ignore us.
Claude: Uses case_study tool
This matches the Steel Mini-Mills pattern closely.
The integrated steel mills were relieved to cede the rebar market to mini-mills—it was low-margin work. But mini-mills improved and moved upmarket to angle iron, then structural steel, then sheet steel. At each tier, the integrated mills rationally retreated.
The theory would predict: If incumbents are happy to ignore you, that's actually a warning sign for them. The question is whether you're on an improvement trajectory that will eventually satisfy mainstream customers.
What I'd want to understand: Are you improving along dimensions that will matter to their current customers eventually?
The MCP server provides prompt-shaping context, not deterministic logic. When you call a tool:
This design lets Claude do what it does best—nuanced reasoning and judgment—while the MCP tools ensure the analysis stays grounded in Christensen's theory.
The agent is designed to maintain high fidelity to Christensen's actual thinking:
Must include:
Must avoid:
Run npm test to validate outputs against these fidelity markers.
# Build TypeScript
npm run build
# Watch mode
npm run dev
# Run validation tests
npm test
# Test with MCP Inspector
npm run inspector
christensen-mcp/
├── src/
│ ├── index.ts # MCP server entry point
│ ├── persona-loader.ts # YAML persona parser
│ ├── personas/
│ │ └── christensen.yaml # Core persona definition
│ ├── frameworks/
│ │ ├── types.ts # Shared types
│ │ ├── jobs-to-be-done.ts # JTBD framework
│ │ ├── disruption.ts # Disruption theory
│ │ └── capabilities.ts # CPP + Resource Dependence
│ ├── tools/
│ │ ├── analyze-decision.ts # Main analysis tool
│ │ ├── case-study.ts # Case study explorer
│ │ └── get-framework.ts # Framework reference
│ └── validation/
│ ├── fidelity-check.ts # Output validation
│ └── test-decisions.ts # Test cases
├── docs/
│ ├── BUILD_LOG.md # Development process journal
│ └── CHRISTENSEN_SPEC.md # Persona specification
└── dist/ # Compiled JavaScript
This agent was built as a proof-of-concept for a persona agent factory. The BUILD_LOG.md documents the entire process so future agents can be built faster.
Key patterns extracted:
Built on Clayton Christensen's work, including:
This is an educational tool meant to help apply Christensen's publicly available frameworks. It is not affiliated with or endorsed by the Christensen Institute.
MIT
Built with Claude Code as a demonstration of MCP server development and persona agent architecture.
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent