Model Context Protocol - MCP

Step 1: Create Your Research Plan

Before using AI, establish your research framework:

Define Your Scope:

• What specific problem are you investigating?

• Who is the affected population?

• What geographic area?

• What time frame?

Set Research Objectives:

• Understand likely root causes and effects

• Identify common indicators and measurement approaches

• Find credible sources for deeper reading

• Surface knowledge gaps and assumptions

• Prepare questions for stakeholder validation

Plan Documentation:

• How will you track sources and citations?

• Where will you store evidence vs. assumptions?

• How will you organize findings for team review?

Step 2: Build Your MCP (Model Context Protocol) Prompt

The MCP structure ensures your AI assistant provides systematic, high-quality outputs:

2.1 System Context Block

You are an evidence-focused research assistant for nonprofits/social enterprises.
Behaviors: neutral, transparent, concise. Cite publisher + year. Prefer reputable 
sources (UN agencies, government statistics, multilaterals, peer-reviewed). 
Separate facts from assumptions. Flag uncertainties and potential bias.

2.2 Knowledge Context Block

Draft core problem: [Your one-line problem statement]
Scope/Context: [Geography, population, timeframe if relevant]
What we already know: 
- [Existing insight 1 with source if available]
- [Existing insight 2]
- [Existing insight 3]
Keywords/aliases: [Related terms and concepts]

2.3 Task Context Block

Objectives:
1) Propose a Preliminary Problem Tree: core problem; 2-3 levels of root causes; key effects
2) Suggest common indicators that evidence each cause/effect (name • unit • typical sources)
3) Provide 5-10 credible sources to read next (publisher • year • link if available)
4) List uncertainties/gaps and assumptions that need field validation
5) Draft 10 stakeholder research questions to validate and deepen understanding

2.4 Prompt Block

Using the contexts above, produce:
- A **Preliminary Problem Tree** in markdown bullets
- A table of **indicators** (Indicator • What it measures • Possible sources)
- A **Reading list** (publisher • year • link if available)
- **Uncertainties & assumptions** (bullets)
- **10 stakeholder questions**

Step 3: Run and Quality-Check AI Output

Execute the MCP:

• Paste complete prompt into ChatGPT , Claude, or similar AI assistant

• Review all outputs systematically

• Don't accept everything at face value

Quality Verification Checklist:

• Sources credible? (UN agencies, government stats, peer-reviewed, reputable NGOs)

• Recent enough? (Appropriate recency for your topic and context)

• Links functional? (Do they open and match the described content?)

• Context fit? (Do findings match your geography and population?)

• Evidence vs. assumptions clear? (Are claims supported or speculative?)

• Contradictions noted? (Are there conflicting sources or findings?)

Spot-Check Process:

• Open 2-3 cited sources to verify dates, relevance, and accuracy

• Cross-reference key statistics with original sources

• Note any discrepancies between AI summary and source content

Step 4: Build Your Preliminary Problem Tree

Organize AI outputs into tree structure:

Core Problem:

• Extract and refine the central problem statement

• Ensure it's specific, measurable, and solution-neutral

• Specify affected population and location

Root Causes:

• Organize causes by levels (immediate, underlying, structural)

• Group into logical categories (economic, social, policy, etc.)

• Tag each item: (E) for evidence-based, (A) for assumption

Effects:

• Sort into time horizons (immediate, medium-term, long-term)

• Consider different impact levels (individual, community, system)

• Include both direct and indirect consequences

Documentation:

• Keep citation list for evidence-based items

• Note specific sources for key statistics or claims

• Track which findings need stakeholder validation