Stakeholder Map

Research and map the full connection network for Youthfront stakeholders: $ARGUMENTS

$ARGUMENTS should be one of:

• A stakeholder name (e.g. Tim Smith) — research a single stakeholder
• --all — research all stakeholders in the registry
• --unresearched — research only stakeholders without an existing connection map
• --refresh NAME — re-research a previously mapped stakeholder
• --add "Name" "Role" "Title" — add a new stakeholder to the registry, then research them
• Empty — show the current registry and ask the user which stakeholder(s) to research

Context

This skill is used outside the Youthfront donor platform to seed stakeholder connection data that will later be imported into the app. "Stakeholders" are the people in Youthfront's inner orbit — board members, executives, key staff, major donors, church partners — whose existing relationships are the bridges to new high-value donors.

The goal is to produce a comprehensive, structured map of each stakeholder's external connections: who they know, through what context, and how those connections could become pathways to donor cultivation.

Project path: ~/Desktop/Dev/repos/non-profit-dashboard

Before Starting

1. Read the stakeholder registry: ~/Desktop/Dev/repos/non-profit-dashboard/public/data/insiders.json
2. Read existing LinkedIn data: ~/Desktop/Dev/repos/non-profit-dashboard/public/data/insiders-linkedin.json
3. Read any existing connection maps in: ~/Desktop/Dev/repos/non-profit-dashboard/public/data/stakeholder-connections/
4. Check LinkedIn MCP session health — run a lightweight lookup to confirm the session is active. If it fails, tell the user to run uvx linkedin-scraper-mcp --login to re-authenticate.
5. Read the connection mapping strategy doc for context on relationship types and scoring: ~/Desktop/Dev/repos/non-profit-dashboard/_docs/agentic-connection-mapping.md

If the stakeholder-connections/ directory doesn't exist, create it.

Research Pipeline

For each stakeholder, execute these stages sequentially. Stream progress to the user after each stage.

Stage 1 — Compile Existing Intelligence

Gather everything already known about this stakeholder from project files:

1. Pull their record from insiders.json (affiliations, role, summary, sources)
2. Pull their LinkedIn enrichment from insiders-linkedin.json (experience, education, skills, connections)
3. Check if any prospect records reference them (search prospects-tier1.json through prospects-tier4-part2.json for name mentions in fields.network_connections or similar)
4. Check harvest data if available: _docs/api-sources/kc-donor-prospecting/harvest/output/ for LittleSis relationships, 990 cross-references, FEC co-giving

Compile into a working profile document. Identify gaps — what affiliations, boards, churches, clubs, or networks are unknown or unconfirmed.

Stage 2 — LinkedIn Deep Dive

Use the LinkedIn MCP tools to pull the stakeholder's full profile and extract connection signals:

mcp__linkedin__get_person_profile:
  url: [stakeholder's LinkedIn URL from registry]
  sections: experience, education, interests, honors, contact_info

From the profile, extract and structure:

• Current and past employers — with titles, dates, and overlap potential
• Board and volunteer roles — explicitly listed or mentioned in descriptions
• Education — alma maters, graduation years, honors societies
• Professional associations — any mentioned in headline, about, or experience
• Publications and speaking — books, articles, conference appearances
• Endorsements and skills — top skills indicate professional circles

Then search for adjacent stakeholders at the same organizations:

mcp__linkedin__search_people:
  keywords: [organization name]
  company: [organization name]
  location: "Kansas City"

Run this for the stakeholder's top 3-5 most connection-rich affiliations (boards, employers, churches). The goal is to find who else is in those organizations — especially people who might also appear in the prospect database.

Rate limit awareness: Track lookups. Stop LinkedIn searches if approaching 80 for the day. Report remaining quota to the user.

Stage 3 — Web Research

Run targeted web searches to fill gaps LinkedIn doesn't cover:

1. Board memberships: "{stakeholder name}" board OR trustee OR director "Kansas City"
2. Church affiliation: "{stakeholder name}" church OR congregation OR parish "Kansas City"
3. Civic and social: "{stakeholder name}" rotary OR chamber OR country club OR "Kansas City"
4. Philanthropy: "{stakeholder name}" foundation OR gift OR donor OR fundraiser "Kansas City"
5. Event co-attendance: "{stakeholder name}" gala OR luncheon OR awards "Kansas City" 2024 OR 2025 OR 2026
6. Family connections: "{stakeholder name}" spouse OR wife OR husband "Kansas City" (only if relevant signals exist)
7. News and features: "{stakeholder name}" Kansas City Business Journal OR Startland OR Flatland

For each result, extract:

• The organization or context (e.g., "Nelson-Atkins Museum Board of Trustees")
• The relationship type (board, church, professional, social, civic, philanthropic, family, event)
• The evidence (URL or description)
• Whether the affiliation is current or historical

Stage 4 — Cross-Reference Against Prospect Database

This is where connection mapping happens. For every affiliation discovered in Stages 1-3:

1. Search the prospect database for others at the same organization:
   • Grep prospect JSON files for the organization name
   • Check fields.board_roles, fields.employer, fields.church_affiliations, fields.network_connections
2. Search the stakeholder registry for other insiders at the same organization
3. For each match, record a connection edge

Stage 5 — Synthesize Connection Map

Compile all discoveries into a structured connection map for this stakeholder.

Output Format

For each stakeholder, write a JSON file to: ~/Desktop/Dev/repos/non-profit-dashboard/public/data/stakeholder-connections/{stakeholder-id}.json

{
  "stakeholder_id": "ins-tim-smith",
  "stakeholder_name": "Tim Smith",
  "role": "board",
  "title": "Board Chair",
  "researched_at": "2026-03-22T14:30:00Z",
  "research_sources": ["linkedin", "web_search", "990_crossref", "fec_cogiving", "littlesis"],
  "linkedin_quota_used": 3,

  "profile": {
    "summary": "Board chair with 30+ years NPO fundraising leadership...",
    "location": "Kansas City, MO",
    "current_employer": "Non Profit DNA",
    "church": "Village Presbyterian Church",
    "neighborhood": "Prairie Village",
    "alma_maters": ["Liberty University"],
    "professional_focus": ["nonprofit fundraising", "faith-based development"]
  },

  "affiliations": [
    {
      "organization": "Non Profit DNA",
      "type": "employer",
      "role": "President",
      "status": "active",
      "since": "2000",
      "source": "linkedin",
      "connection_potential": "high",
      "notes": "Consulting practice — clients are NPO leaders, many likely in prospect DB"
    },
    {
      "organization": "Museum of the Bible",
      "type": "employer",
      "role": "Former CDO",
      "status": "historical",
      "period": "2014-2017",
      "source": "linkedin",
      "connection_potential": "moderate",
      "notes": "DC-based but faith/museum donor network may overlap KC"
    }
  ],

  "connections": [
    {
      "connected_to": {
        "id": "prospect-uuid-here",
        "name": "James Hebenstreit",
        "type": "prospect",
        "tier": 1
      },
      "connection_type": "board",
      "shared_context": "Both serve on Nelson-Atkins Museum Board of Trustees",
      "strength": "strong",
      "evidence": "Nelson-Atkins 2025 annual report",
      "evidence_url": "https://example.com/source",
      "status": "active",
      "discovered_via": "web_search",
      "approach_suggestion": "Tim mentions Youthfront's camp impact at the next Nelson-Atkins board dinner — natural peer conversation between fellow trustees."
    },
    {
      "connected_to": {
        "id": "ins-mike-king",
        "name": "Mike King",
        "type": "stakeholder",
        "role": "executive"
      },
      "connection_type": "professional",
      "shared_context": "Both former Youthfront executives; Tim was VP Development 1996-2000",
      "strength": "strong",
      "evidence": "LinkedIn employment history",
      "status": "active",
      "discovered_via": "linkedin"
    }
  ],

  "connection_summary": {
    "total_connections": 12,
    "prospect_connections": 7,
    "stakeholder_connections": 5,
    "by_type": {
      "board": 3,
      "professional": 4,
      "church": 2,
      "philanthropic": 2,
      "social": 1
    },
    "strongest_bridges": [
      "Nelson-Atkins Museum Board (3 prospects)",
      "Village Presbyterian Church (2 prospects)",
      "Non Profit DNA client network (2 prospects)"
    ]
  },

  "gaps_and_recommendations": [
    "Church affiliation unconfirmed — ask Tim directly",
    "Country club / social memberships unknown — high-value gap for Leawood/Mission Hills prospects",
    "Children's schools unknown — could unlock parent network connections",
    "Tim's consulting clients (Non Profit DNA) likely overlap with prospect DB but are not publicly listed"
  ],

  "new_stakeholder_candidates": [
    {
      "name": "Sarah Chen",
      "reason": "Co-trustee at Nelson-Atkins, appears connected to 4 Tier 1 prospects",
      "recommended_action": "Add to stakeholder registry and research"
    }
  ]
}

After Each Stakeholder

1. Update insiders.json — fill in any new affiliations, church, or location data discovered
2. Update insiders-linkedin.json — if new LinkedIn data was pulled
3. Write the connection map JSON file
4. Report to the user:
   • Stakeholder name and role
   • Number of connections discovered (by type)
   • Top 3 strongest bridge opportunities
   • Any new stakeholder candidates discovered
   • Gaps that need human input
   • LinkedIn quota remaining

After All Stakeholders (batch mode)

Write a master graph file to: ~/Desktop/Dev/repos/non-profit-dashboard/public/data/stakeholder-connections/_graph.json

{
  "generated_at": "2026-03-22T16:00:00Z",
  "stakeholder_count": 17,
  "total_connections": 89,
  "total_unique_prospects_reached": 42,

  "edges": [
    {
      "from": "ins-tim-smith",
      "to": "prospect-uuid",
      "connection_type": "board",
      "shared_context": "Nelson-Atkins Museum Board",
      "strength": "strong"
    }
  ],

  "hub_organizations": [
    {
      "organization": "Nelson-Atkins Museum",
      "stakeholders_connected": ["ins-tim-smith", "ins-cheryl-reinhardt"],
      "prospects_connected": ["uuid-1", "uuid-2", "uuid-3"],
      "total_people": 5,
      "connection_type": "board"
    }
  ],

  "multi_path_prospects": [
    {
      "prospect_id": "prospect-uuid",
      "prospect_name": "James Hebenstreit",
      "paths": 3,
      "connectors": ["ins-tim-smith", "ins-ed-garlich", "ins-mike-king"],
      "strongest_path_score": 87
    }
  ],

  "new_stakeholder_candidates": [
    {
      "name": "Sarah Chen",
      "nominated_by": ["ins-tim-smith", "ins-cheryl-reinhardt"],
      "reason": "Appears in 2 stakeholder connection maps, connected to 4 Tier 1 prospects",
      "recommended_priority": "high"
    }
  ],

  "coverage_gaps": [
    "3 board members have no LinkedIn profile — manual research needed",
    "Church affiliations confirmed for only 4 of 17 stakeholders",
    "No stakeholder coverage in healthcare/hospital sector"
  ]
}

Key Design Rules

• Incremental — Skip stakeholders who already have a connection map file unless --refresh is used. Always check for existing data before researching.
• Source everything — Every connection must have a discovered_via field and evidence. No hallucinated connections.
• LinkedIn budget — Track daily usage. Never exceed 80 lookups. Report quota after each stakeholder.
• Prospect matching — When cross-referencing, use fuzzy matching (case-insensitive, handle middle names/initials). Flag uncertain matches for human review.
• Stakeholder discovery — If research reveals someone who is clearly a "hidden stakeholder" (e.g., a major donor's spouse who sits on 3 boards with Tier 1 prospects), add them to new_stakeholder_candidates — do NOT auto-add to the registry without user confirmation.
• Privacy — Do not record personal phone numbers, home addresses, or non-public email addresses. Stick to professional affiliations and publicly available information.
• Idempotent writes — Connection map files can be safely overwritten. The researched_at timestamp indicates freshness.
• Confirm before batch — If --all is used and there are more than 5 unresearched stakeholders, show the list and estimated LinkedIn quota cost before proceeding. Ask the user to confirm.

Error Handling

• LinkedIn session expired — Stop and instruct user to run uvx linkedin-scraper-mcp --login
• LinkedIn rate limit approaching — Warn at 60 lookups, hard stop at 80. Save progress and report what was completed.
• Stakeholder not found in registry — Suggest using --add to register them first
• No prospect matches found — This is a valid outcome. Record it in the connection map with empty connections array and note in gaps_and_recommendations.
• Ambiguous prospect match — Record the match with "confidence": "low" and flag for human review

Progress Reporting

After completing research for each stakeholder, output a summary block:

## Stakeholder Research Complete: Tim Smith (Board Chair)

### Connections Discovered: 12
- Board co-service: 3 (Nelson-Atkins, Kauffman Foundation, UMKC Trustees)
- Professional overlap: 4 (Non Profit DNA clients, Museum of the Bible alumni)
- Church: 2 (Village Presbyterian)
- Philanthropic: 2 (KC Community Foundation, Midwest Trust)
- Social: 1 (Mission Hills Country Club — unconfirmed)

### Top Bridge Opportunities
1. **James Hebenstreit** via Nelson-Atkins Board (strength: strong)
2. **Margaret Sullivan** via Village Presbyterian (strength: moderate)
3. **David Chen** via Kauffman Foundation (strength: moderate)

### New Stakeholder Candidates
- Sarah Chen — co-trustee at Nelson-Atkins, connected to 4 Tier 1 prospects

### Gaps Needing Human Input
- Country club membership unconfirmed
- Children's schools unknown
- Non Profit DNA client list not publicly available

### LinkedIn Quota: 74/100 remaining today