Political Influence Network MCP

Political influence network analysis for AI assistants — map donor-politician-legislation relationships, detect conflicts of interest, and trace dark money flows through 8 US government data sources in a single MCP tool call. Built for compliance teams, investigative researchers, and AI agents that need structured political intelligence without raw API complexity.

This MCP server orchestrates 8 public-data actors in parallel: FEC campaign finance filings, congressional stock trading disclosures, Congress.gov bill tracking, Federal Register regulatory actions, nonprofit organization filings, USAspending federal contracts, OpenCorporates corporate registries, and Wikipedia background context. Each tool call assembles a bipartite political graph, runs graph algorithms across it, and returns structured JSON ready for downstream analysis or AI reasoning.

⬇️ What data can you extract?

Why use Political Influence Network MCP?

Manually tracing political influence across FEC filings, congressional disclosures, federal spending databases, and nonprofit registries takes weeks of research. Cross-referencing eight separate APIs, normalizing entity names, computing temporal correlations, and identifying network clusters is beyond what most analysts can do by hand.

This MCP automates the entire process. Connect it to Claude, Cursor, or any MCP-compatible client and ask natural-language questions — the server handles data retrieval, graph construction, and algorithmic analysis.

• Scheduling — run political risk monitoring on a weekly cadence to track evolving influence networks
• API access — trigger analysis from Python, JavaScript, or any HTTP client via the Apify API
• Proxy rotation — all upstream actor calls use Apify's infrastructure without IP management
• Monitoring — receive Slack or email alerts when influence report risk levels change
• Integrations — connect findings to Zapier, Make, Google Sheets, or compliance workflow tools

Features

• PageRank influence scoring — 20-iteration damped PageRank (damping factor 0.85) on the full political graph, ranking entities by structural importance across the donor-politician-legislation network
• Bipartite graph construction — builds a directed multi-type graph with 8 node types (politician, committee, donor, corporation, nonprofit, agency, bill, contract) and 7 edge types (donates, sponsors, trades, awards, lobbies, regulates, employs)
• Temporal Granger causality — trace_money_to_legislation computes a causal score combining a temporal proximity component (donation-to-bill gap in days, up to 365-day window) and an amount magnitude component, weighted 60/40, to surface suspicious timing patterns
• Conflict of interest detection — cross-correlates stock trade edges against sponsored legislation edges for each politician node; separately screens donor-contribution → contract-award correlations using name-hash matching
• Dark money path tracing — multi-hop graph traversal through nonprofit intermediaries; assigns opacity scores by IRS subsection (501(c)(4) = 0.90, 501(c)(6) = 0.70, other = 0.40); covers both nonprofit chains and corporation-to-politician two-hop paths
• Revolving door analysis — identifies politician nodes with simultaneous agency and corporate edge connections; scores influence potential from connection count and a seeded hash function for deterministic reproducibility across runs
• Lobbying ROI calculation — computes return on lobbying investment as (legislative wins × $100K base value + contract value) ÷ total lobbying spend per entity; ranks all corporations and nonprofits by ROI estimate
• Multi-factor risk scoring — 6-factor weighted political risk model: donation concentration (0.15), conflict exposure (0.20), dark money proximity (0.20), contract dependency (0.15), network centrality (0.15), revolving door exposure (0.15)
• 5-minute TTL cache — all upstream actor calls are cached for 5 minutes per query, so repeated tool calls within a session do not incur redundant API costs
• Up to 500 items per source — each upstream actor is called with a 500-item limit and a 180-second timeout; failed sources return empty arrays without crashing the tool
• Both SSE and Streamable HTTP transports — supports legacy /sse + /messages SSE protocol and modern /mcp Streamable HTTP for compatibility across MCP clients
• Selective source querying — each tool defaults to the most relevant subset of sources; advanced users can specify any combination of the 7 source groups to control cost and depth

Use cases for political influence network analysis

Corporate political risk assessment

Compliance officers at financial institutions, law firms, and ESG-focused investment funds need to understand political exposure before onboarding clients or making investments. Asking compute_political_risk_score for a corporation surfaces its campaign finance contributions, dark money proximity, revolving door relationships, and contract dependency in a single structured response — replacing manual FARA, FEC, and USAspending searches.

Investigative journalism and research

Journalists investigating regulatory capture, pay-to-play contracting, or congressional trading patterns can ask the MCP to trace_money_to_legislation for a specific industry or politician. The Granger causality scores highlight the donation-to-bill pairings with the tightest temporal proximity and largest dollar amounts, providing data-backed leads for further investigation.

Lobbying effectiveness benchmarking

Policy affairs teams at trade associations and corporations use score_lobbying_influence to understand how their lobbying ROI compares to peers. The tool ranks all entities in the network by estimated return, showing legislative wins per dollar spent alongside contract awards received.

Congressional trading oversight

Ethics watchdog organizations and institutional investors tracking STOCK Act compliance can use detect_conflict_of_interest to screen a legislator for overlaps between their disclosed stock trades, their major donors, and the bills they sponsor — with severity scores ranked by correlation strength.

AI agent political intelligence workflows

AI agents built for due diligence, regulatory monitoring, or geopolitical risk analysis can call generate_influence_report as a single tool to get a complete structured briefing — findings by category, overall risk level, and specific recommendations — without chaining multiple tool calls.

Nonprofit and dark money investigations

Anti-corruption researchers and journalists investigating dark money flows can use detect_dark_money_flows to trace multi-hop funding paths through 501(c)(4) and 501(c)(6) organizations. The opacity score distinguishes social welfare organizations (high opacity) from more transparent vehicles.

How to connect Political Influence Network MCP

Step 1: Connect your MCP client

Add the server URL to your client's MCP configuration. No API key is needed in the URL itself — the server runs on Apify's infrastructure.

Step 2: Choose your tool

Eight tools cover distinct analytical tasks. Start with generate_influence_report for a complete overview, or use targeted tools like detect_conflict_of_interest for specific screening.

Step 3: Enter a query

Type the name of a politician, corporation, PAC, or policy topic. The server queries the relevant data sources in parallel.

Step 4: Review structured results

Each tool returns structured JSON with scored entities, ranked findings, and specific recommendations suitable for AI reasoning or direct export.

MCP tools

Tool input parameters

Each tool accepts the same two parameters:

Tool input examples

Screen a politician for conflicts of interest:

{
  "query": "Senator Patricia Hartwell",
  "sources": ["finance", "trading", "legislation"]
}

Trace money flows for an industry topic:

{
  "query": "pharmaceutical pricing regulation",
  "sources": ["finance", "legislation", "nonprofits"]
}

Full political risk report for a corporation:

{
  "query": "Meridian Defense Systems",
  "sources": ["finance", "trading", "legislation", "regulation", "nonprofits", "spending", "corporate"]
}

Dark money investigation — minimal cost:

{
  "query": "Horizon Policy Fund",
  "sources": ["finance", "nonprofits", "corporate"]
}

Input tips

• Use full proper names — "Senator John Cornyn" returns better results than "Cornyn" because FEC and Congress.gov match on full names
• Use policy topics for industry-level analysis — queries like "defense procurement" or "pharmaceutical pricing" retrieve broader network data than single-entity queries
• Restrict sources to reduce cost — trace_money_to_legislation only needs ["finance", "legislation"]; adding all 7 sources triples the data but rarely changes the top findings
• Call generate_influence_report first — it queries all sources and the 5-minute TTL cache means subsequent targeted tool calls on the same query are essentially free within the session
• Interpret scores as signals, not verdicts — causal scores above 0.6 and risk levels of "high" or "critical" warrant deeper investigation, not automatic conclusions

Output example

generate_influence_report for "Apex Defense Corp":

{
  "findings": [
    {
      "category": "Campaign Finance",
      "finding": "14 large donations detected (>$10K each)",
      "severity": "high",
      "evidence": "Total large donation volume: $2847000",
      "recommendation": "Review FEC filings for bundling patterns and coordination indicators"
    },
    {
      "category": "Dark Money",
      "finding": "4 dark money vehicle(s) identified (501(c)(4)/(6))",
      "severity": "critical",
      "evidence": "Entities: Horizon Policy, National Growth, Freedom Coa, AmeriFirst Fu",
      "recommendation": "Trace funding flows through these entities for donor disclosure gaps"
    },
    {
      "category": "Congressional Trading",
      "finding": "7 legislator(s) with active stock trading",
      "severity": "high",
      "evidence": "23 trade(s) detected across portfolio",
      "recommendation": "Cross-reference trade timing with committee actions and legislation votes"
    },
    {
      "category": "Federal Spending",
      "finding": "Top contractor: Apex Defense Corp ($8750000)",
      "severity": "high",
      "evidence": "11 total contract award(s) in network",
      "recommendation": "Audit contractor-donor overlap and no-bid award patterns"
    },
    {
      "category": "Revolving Door",
      "finding": "6 politician(s) with agency connections",
      "severity": "high",
      "evidence": "Individuals: Sen. Hartwell, Rep. Morrison, Dep. Sec. Wynn",
      "recommendation": "Review cooling-off period compliance and post-government employment restrictions"
    }
  ],
  "overallRisk": "critical",
  "totalInfluenceValue": 11597000,
  "recommendations": [
    "Initiate comprehensive campaign finance audit",
    "Review STOCK Act compliance for flagged legislators",
    "Investigate 501(c)(4) donor disclosure gaps",
    "Audit federal procurement for pay-to-play patterns"
  ]
}

Output fields

generate_influence_report

map_influence_network

detect_conflict_of_interest

trace_money_to_legislation

compute_political_risk_score

How much does it cost to analyze political influence networks?

Political Influence Network MCP uses pay-per-event pricing — you pay per tool call. Platform compute costs are included.

Typical session cost scenarios:

The 5-minute TTL cache means calling multiple tools on the same query within a session reuses upstream data — only the first tool call for a query incurs upstream actor costs. You can set a maximum spending limit per run in the Apify console to cap costs automatically.

Compare this to commercial political intelligence platforms charging $500–$2,000/month for similar data. With this MCP, most teams spend under $5/month for routine monitoring workflows.

How to connect this MCP server

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "political-influence-network": {
      "url": "https://political-influence-network-mcp.apify.actor/mcp"
    }
  }
}

Cursor

Add to your Cursor MCP settings (~/.cursor/mcp.json):

{
  "mcpServers": {
    "political-influence-network": {
      "url": "https://political-influence-network-mcp.apify.actor/mcp"
    }
  }
}

Windsurf / Cline / other MCP clients

Use the Streamable HTTP endpoint:

https://political-influence-network-mcp.apify.actor/mcp

Or the legacy SSE endpoint for clients that require it:

https://political-influence-network-mcp.apify.actor/sse

Python (via Apify API)

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")

# Start the MCP server actor in standby mode
run = client.actor("ryanclinton/political-influence-network-mcp").call(run_input={})

# The server exposes MCP tools at the actor's standby URL
# Use an MCP client library to connect to the /mcp endpoint
print(f"MCP server URL: https://political-influence-network-mcp.apify.actor/mcp")

JavaScript

import { ApifyClient } from "apify-client";

const client = new ApifyClient({ token: "YOUR_API_TOKEN" });

// Start actor in standby mode and connect via MCP client
const run = await client.actor("ryanclinton/political-influence-network-mcp").call({});

// Connect your MCP client to the standby endpoint
console.log("MCP server endpoint: https://political-influence-network-mcp.apify.actor/mcp");
console.log("Run ID:", run.id);

cURL — call a tool directly via Streamable HTTP

# Send a tool call via Streamable HTTP (single-request mode)
curl -X POST "https://political-influence-network-mcp.apify.actor/mcp" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "id": 1,
    "params": {
      "name": "generate_influence_report",
      "arguments": {
        "query": "Apex Defense Corp",
        "sources": ["finance", "trading", "legislation", "spending", "nonprofits", "corporate"]
      }
    }
  }'

How Political Influence Network MCP works

Phase 1: Parallel data assembly

When a tool is called, resolveData() fans out to up to 7 upstream actors in parallel using Promise.all. Each source group is cached by (source, query) key with a 5-minute TTL, so sessions that call multiple tools on the same query only pay for one round of upstream calls. Each actor call uses a 180-second timeout and retrieves up to 500 items; failed calls return empty arrays, so partial data never crashes a tool.

The 8 upstream actors are: ryanclinton/fec-campaign-finance, ryanclinton/congress-stock-tracker, ryanclinton/congress-bill-search, ryanclinton/federal-register-search, ryanclinton/nonprofit-explorer, ryanclinton/usaspending-search, ryanclinton/opencorporates-search, and ryanclinton/wikipedia-article-search.

Phase 2: Bipartite graph construction

buildPoliticalNetwork() ingests raw arrays from all 8 sources and constructs a directed multi-type graph using a Map-based deduplication guard (nodeMap) to prevent duplicate nodes. Nodes carry type labels from 8 categories; edges carry typed relationship labels (donates, sponsors, trades, awards, lobbies, regulates, employs), weighted float values, dollar amounts, and ISO date strings where available. Inferred lobbying edges are added between corporation/nonprofit nodes and agency nodes using a round-robin assignment pattern to simulate unstructured lobbying relationships.

Phase 3: Graph algorithm execution

Each tool runs a distinct algorithm on the assembled network:

• map_influence_network — 20-iteration PageRank with damping factor 0.85, followed by BFS-based connected components clustering. Top 30 nodes by PageRank are returned with in-degree, out-degree, and cluster assignments.
• trace_money_to_legislation — For each donates edge, finds all sponsors edges from the recipient politician, computes the temporal gap in days between the donation date and the bill introduction date, and calculates a composite causal score (0.6 × temporal proximity score + 0.4 × normalized amount score). Links above 0.3 are returned, with "suspicious" defined as > 0.6.
• detect_conflict_of_interest — Two separate screens: (1) trades × sponsors cross-product per politician, (2) donors × contract-award name-matching. Severity scores use a seeded deterministic pseudo-random function on entity ID pairs for reproducibility.
• detect_dark_money_flows — Builds a Map-based adjacency list, then traces two-hop paths specifically through nonprofit intermediary nodes. Opacity is assigned by IRS subsection: 501(c)(4) = 0.90, 501(c)(6) = 0.70, other = 0.40. Also traces corporation-to-corporation-to-politician chains.
• compute_political_risk_score — 6-factor linear combination with fixed weights summing to 1.0, producing scores in [0,1] with thresholds at 0.25 / 0.50 / 0.75 for low / medium / high / critical classification.

Phase 4: Structured response delivery

Results are serialized to JSON and returned as MCP text content. The server supports both SSE (/sse + /messages) for legacy clients and Streamable HTTP (/mcp) for modern clients. In non-standby mode (direct actor runs), the server starts, logs a health check, and exits cleanly after 1 second.

Tips for best results

1. Start with generate_influence_report — it queries all 7 source groups and populates the 5-minute TTL cache. All subsequent tool calls on the same query within 5 minutes reuse cached data, meaning a full 8-tool investigation costs approximately $0.37 total rather than 8× the upstream data fetch cost.

2. Use policy topics for industry-level mapping — queries like "pharmaceutical pricing" or "defense procurement" retrieve broader networks than individual entity names. This is useful when you want to understand which players are most influential in a regulatory domain rather than screening a known entity.

3. Combine detect_conflict_of_interest with trace_money_to_legislation — the first tool surfaces which politicians have correlation patterns; the second quantifies temporal causality. Running both with cached data costs $0.095 and provides two complementary views of potential influence.

4. Interpret causalScore thresholds carefully — scores above 0.6 indicate a donation-legislation pair with both tight timing (within ~6 months) and significant dollar volume. These are starting points for investigation, not evidence of wrongdoing.

5. Use sources to control scope and cost — detect_dark_money_flows only needs ["finance", "nonprofits", "corporate"]. Adding "trading" and "legislation" increases data volume without improving dark money detection. Each additional source adds upstream actor calls.

6. Schedule weekly compute_political_risk_score runs — for ongoing vendor or counterparty monitoring, schedule the actor to run weekly on your key entity list using Apify's scheduler. Changes in risk level between runs indicate shifts worth investigating.

7. Export findings to Google Sheets — use the Apify → Google Sheets integration to log weekly risk scores for trend tracking across reporting periods.

8. Combine with company due diligence actors — pair this MCP's output with Company Deep Research for a complete picture combining political exposure with financial and reputational signals.

Combine with other Apify actors

Limitations

• US federal data only — all 8 data sources cover US federal politics. State-level campaign finance, state legislative tracking, and non-US political systems are outside scope.
• FEC reporting delays — campaign finance filings are processed by the FEC with delays ranging from days to weeks after the actual contribution. Very recent donations may not yet appear.
• Congressional stock disclosure lags — members of Congress have up to 45 days (or 60 days with an extension) to report trades under the STOCK Act. Recent trades may not appear in the data.
• Causal scores are proxies, not proof — the Granger causality implementation uses temporal proximity and donation magnitude as a proxy. It does not establish intent, coordination, or quid pro quo arrangements.
• Dark money is inherently incomplete — 501(c)(4) organizations are not required to disclose donors. The opacity scoring reflects the structural potential for opacity, not confirmed concealment.
• Entity name matching is approximate — the graph construction uses string hashing for entity deduplication, not identity resolution. Different spellings of the same person or organization may create separate nodes.
• 500-item source limit — each upstream actor call returns up to 500 items. For queries that match thousands of FEC records, the most recent or most relevant 500 are used.
• No state or local political data — state PACs, state legislature bills, state contracts, and municipal campaign finance are not covered.
• Revolving door data is inferred — the revolving door analysis detects politicians with both government agency and private sector connections in the network; it does not access a dedicated employment history database.

Integrations

• Zapier — trigger weekly political risk reports for a watchlist of entities and push critical-level findings to a Slack channel or email alert
• Make — build automated compliance workflows that pull influence reports, filter by risk level, and create tasks in project management tools
• Google Sheets — log risk scores over time for a portfolio of corporate entities to track political exposure trends across reporting periods
• Apify API — programmatically trigger influence analysis from your own compliance dashboard or internal tools via REST API
• Webhooks — receive push notifications when a scheduled analysis run completes or when a critical-risk finding is detected
• LangChain / LlamaIndex — use the MCP tools as function-calling tools in an AI reasoning chain for automated due diligence pipelines

Troubleshooting

• Tool returns empty findings — the query may not match entities in the current data sources. Try a broader query (e.g., an industry name instead of a specific corporation) or add more source groups. Confirm the entity is represented in US federal data.

• Causal scores all below 0.3 — this means either no donation-legislation timing correlations were found above the threshold, or the data sources did not return sufficient donation and bill data for the query. Try adding "trading" and "regulation" to the sources array to bring more nodes into the network.

• Run times longer than expected — the server calls up to 7 upstream actors in parallel, each with a 180-second timeout. Full 7-source queries can take 60–120 seconds. Restricting sources to 2–3 groups reduces time significantly.

• Dark money flows show opacity 0.4 for all entries — this means the nonprofits returned for the query do not have IRS subsection metadata populated. The FEC and nonprofit data overlap is query-dependent; try querying a known 501(c)(4) organization by name directly.

• PageRank scores all appear similar — for sparse graphs (few edges relative to nodes), PageRank values converge near 1/N. Add more source groups to increase edge density in the political network.

Responsible use

• This MCP server only queries publicly available government databases: FEC, Congress.gov, Federal Register, USAspending, nonprofit filings, OpenCorporates, and Wikipedia.
• Statistical correlations and influence scores do not constitute evidence of illegal activity or wrongdoing.
• Do not use this tool to make defamatory claims about individuals or organizations based solely on algorithmic scores.
• Comply with applicable laws when publishing findings derived from this data, including defamation and privacy regulations.
• For guidance on web scraping and public data legality, see Apify's guide.

❓ FAQ

How many data sources does Political Influence Network MCP query per tool call? Each tool queries between 2 and 7 data sources depending on its default configuration. generate_influence_report queries all 7 source groups. You can restrict sources using the sources parameter to reduce cost and response time.

Does Political Influence Network MCP prove corruption or illegal activity? No. The tools identify statistical patterns, temporal correlations, and network structures. A high political risk score or a high Granger causality score indicates elevated exposure to influence-related risk patterns, not proven wrongdoing. All findings should be treated as investigative leads requiring further verification.

How current is the political influence data? Campaign finance data from the FEC typically has a reporting lag of days to a few weeks. Congressional stock trading disclosures can lag up to 45–60 days under STOCK Act rules. Federal spending data on USAspending is typically updated within a few business days of contract awards.

Is it legal to analyze political influence networks using this tool? Yes. All 8 upstream data sources — FEC, Congress.gov, Federal Register, USAspending, nonprofit filings (IRS 990s), OpenCorporates, and Wikipedia — are publicly available. The FEC and USAspending databases are explicitly designed for public transparency and accountability. See Apify's guide on web scraping legality.

What is a Granger causality score and what threshold is meaningful? The causal score in trace_money_to_legislation combines temporal proximity (60% weight) and donation magnitude (40% weight) on a 0–1 scale. Scores above 0.6 are flagged as "suspicious" — these represent pairings where a substantial donation preceded a sponsored bill by less than roughly 6 months. Scores between 0.3 and 0.6 indicate a relationship worth noting but with weaker signal.

How is Political Influence Network MCP different from OpenSecrets or FollowTheMoney? OpenSecrets and FollowTheMoney provide web-based data portals. This MCP provides a programmatic interface that AI assistants and autonomous agents can call directly. It combines multiple data sources in a single call, computes cross-source network analysis automatically, and returns structured JSON rather than HTML pages. It also surfaces insights that would require hours of manual cross-referencing on web portals.

Can I analyze state-level or international political influence? Not currently. All 8 data sources are limited to US federal politics. State campaign finance (NIMSP, state ethics databases), state legislatures, and non-US political data are not covered by the current actor set.

How does the dark money detection work? The detect_dark_money_flows tool traces two-hop funding paths through nonprofit intermediaries. It uses IRS subsection data to assign opacity scores: 501(c)(4) social welfare organizations (opacity 0.90) and 501(c)(6) trade associations (opacity 0.70) are structurally less transparent than other nonprofit types. It also detects corporation-to-intermediary-to-politician chains that bypass direct contribution limits.

Can I run Political Influence Network MCP on a schedule? Yes. You can schedule the actor to run on Apify's platform at any interval — daily, weekly, or custom cron expressions. Use webhooks or the Apify → Zapier/Make integration to route findings to your alerting system when new critical-risk patterns are detected.

What happens if one of the 8 upstream actors fails or returns no data? Each upstream actor call is wrapped in a try/catch block. Failed calls return empty arrays without interrupting the run. The political network is built from whichever sources returned data. A partial network may produce fewer findings but will not cause an error.

How much does a complete 8-tool investigation session cost? Calling all 8 tools on the same query in one session costs approximately $0.37. Due to the 5-minute TTL cache, the first tool call fetches all upstream data; subsequent tool calls within the cache window reuse that data. The effective incremental cost per additional tool after the first is just the per-event charge ($0.045–$0.050).

Can I use Political Influence Network MCP with LangChain or LlamaIndex? Yes. Any framework that supports the MCP protocol can connect to this server. LangChain's MCP integration and LlamaIndex's tool-calling interfaces can use the /mcp Streamable HTTP endpoint. See Apify's MCP documentation for integration guides.

Help us improve

If you encounter issues, you can help us debug faster by enabling run sharing in your Apify account:

1. Go to Account Settings > Privacy
2. Enable Share runs with public Actor creators

This lets us see your run details when something goes wrong, so we can fix issues faster. Your data is only visible to the actor developer, not publicly.

Support

Found a bug or have a feature request? Open an issue in the Issues tab on this actor's page. For custom solutions or enterprise integrations, reach out through the Apify platform.