Skip to main content
Glama

Humanitarian Negotiation MCP Server

OPENAI_LEVERAGE_GUIDE.md5.1 kB
# How to Use Leverage Influence Tool (Tool 4) in OpenAI ## The Problem OpenAI doesn't automatically pass the output from one tool to another. After running **Analyze Stakeholders** (Tool 3), you need to **manually provide that data** to the **Leverage Influence** tool (Tool 4). ## Solution: Copy and Paste Pattern ### Step 1: Run Stakeholder Analysis (Tool 3) Ask Claude to analyze your stakeholders: ``` Analyze these stakeholders: Context: Complex humanitarian crisis requiring coordination between multiple parties... Stakeholders: 1. Riverland Government - Power: 0.9, Urgency: 0.8, Legitimacy: 0.95, Position: 0.6 2. GRN (Rebel Group) - Power: 0.7, Urgency: 0.9, Legitimacy: 0.3, Position: -0.8 3. Local NGOs - Power: 0.5, Urgency: 0.8, Legitimacy: 0.7, Position: 0.7 4. Donors - Power: 0.6, Urgency: 0.7, Legitimacy: 0.8, Position: 0.8 5. Media - Power: 0.4, Urgency: 0.6, Legitimacy: 0.6, Position: 0.5 ``` ### Step 2: Get the Output Claude will return something like: ```json { "analysis_context": "Complex humanitarian crisis...", "total_stakeholders": 5, "stakeholders": [ { "name": "Riverland Government", "power": 0.9, "urgency": 0.8, "legitimacy": 0.95, "position": 0.6, "salience_score": 2.65, "priority_level": "First", "influenced_by": [], "engagement_strategy": "Supportive" }, ...more stakeholders... ], "priority_summary": {...} } ``` ### Step 3: Copy the ENTIRE Response **Important**: You need to copy the COMPLETE JSON response from Tool 3, including: - `analysis_context` - `total_stakeholders` - `stakeholders` (with all fields) - `priority_summary` ### Step 4: Use Leverage Influence Tool Ask Claude to develop tactics for a specific stakeholder, providing the full analysis: ``` Based on the stakeholder analysis I just received, develop influence tactics for the Riverland Government. Here's the complete analysis: [PASTE THE ENTIRE JSON FROM STEP 3 HERE] Target stakeholder: Riverland Government ``` ## Example: Complete Flow ### Input 1 - Analyze Stakeholders: ``` Analyze these 5 stakeholders: Context: Multi-party humanitarian crisis in Riverland requiring coordination. Stakeholders: 1. Riverland Government - Power: 0.9, Urgency: 0.8, Legitimacy: 0.95, Position: 0.6 2. GRN (Armed Group) - Power: 0.7, Urgency: 0.9, Legitimacy: 0.3, Position: -0.8 3. Local NGOs - Power: 0.5, Urgency: 0.8, Legitimacy: 0.7, Position: 0.7 4. International Donors - Power: 0.6, Urgency: 0.7, Legitimacy: 0.8, Position: 0.8 5. Media Networks - Power: 0.4, Urgency: 0.6, Legitimacy: 0.6, Position: 0.5 ``` ### Output 1 - Claude returns stakeholder analysis ```json { "analysis_context": "Multi-party humanitarian crisis in Riverland...", "total_stakeholders": 5, "stakeholders": [ { "name": "Riverland Government", "power": 0.9, "urgency": 0.8, "legitimacy": 0.95, "position": 0.6, "salience_score": 2.65, "priority_level": "First", "influenced_by": [], "engagement_strategy": "Supportive" }, { "name": "GRN (Armed Group)", "power": 0.7, "urgency": 0.9, "legitimacy": 0.3, "position": -0.8, "salience_score": 1.9, "priority_level": "Second", "influenced_by": ["Media Networks"], "engagement_strategy": "Adversarial" }, ... ], "priority_summary": {...} } ``` ### Input 2 - Leverage Influence (with copied data): ``` Now develop influence tactics for Riverland Government using this stakeholder analysis: [PASTE THE COMPLETE JSON HERE] Target: Riverland Government ``` ### Output 2 - Claude develops tactics ```json { "target_stakeholder": "Riverland Government", "target_profile": { "power": 0.9, "urgency": 0.8, "legitimacy": 0.95, "position": 0.6, "priority_level": "First" }, "influence_strategy": { "primary_approach": "Coalition building", "target_psychology": "Leverage shared interests", "communication_tone": "Collaborative" }, "tactical_options": [ { "tactic": "Coalition Building", "description": "Unite with supportive stakeholders...", "allies": ["Local NGOs", "International Donors"] }, ... ] } ``` ## Important Notes 1. **Don't manually edit** the JSON - copy it exactly as Claude returns it 2. **Include all fields** - the stakeholders array must have all properties 3. **One stakeholder at a time** - repeat the process for each stakeholder you want to influence 4. **Keep the structure** - the API expects the exact same structure that Tool 3 returns ## Troubleshooting **Error: "Stakeholder 'X' not found in analysis"** - Make sure the stakeholder name matches exactly (case-sensitive) - Copy the complete JSON response, not a partial one **Error: "Field required" for stakeholders_analysis_json** - You must provide the complete JSON object - Don't just provide the name and target **It says the tool is trying to call but then fails** - Make sure the JSON is valid (use https://jsonlint.com/ to verify) - Check that all required fields are present

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/escenariosparalatransformacion/humanitarian-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server