Module: Intelligence Operations/MCP Intelligence Server Client

Version:

• 2.0.0

Author:

• Hack23 AB - Intelligence Operations Team

License:

• Apache-2.0

Source:

• scripts/mcp-client.js, line 3

See:

• https://riksdag-regering-ai.onrender.com/mcp MCP Server
• https://github.com/Hack23/riksdag-regering-mcp MCP Server Repository
• ./generate-news-enhanced.js News generation using MCP client
• ./data-transformers.js Data transformation pipeline
• docs/MCP_INTEGRATION.md MCP integration guide
• docs/INTELLIGENCE_API_GUIDE.md API reference documentation
• docs/OSINT_COLLECTION.md OSINT collection methodology
• https://www.jsonrpc.org/specification JSON-RPC 2.0 Specification

Examples

Quick Start for GitHub Copilot Agents

// In GitHub Actions workflows with MCP configured, use tools directly:
const events = await mcp["riksdag-regering"].get_calendar_events({
  from: "2026-02-16",
  tom: "2026-02-16",
  limit: 50
});

// Or use this helper client for typed methods:
import MCPClient from './scripts/mcp-client.js';
const client = new MCPClient();
const events = await client.fetchCalendarEvents({ from: today, tom: today });

Handling Cold Starts (30-60 seconds)

// Render.com serverless may have cold starts
console.log("⏳ Warming up MCP server (may take 30-60s)...");

// Start with a simple query to warm up
const status = await client.request('get_sync_status', {});
console.log("✅ MCP server ready");

// Now batch your main queries (will be fast)
const [events, votes, docs] = await Promise.all([
  client.fetchCalendarEvents({ from: today, tom: today }),
  client.fetchVotingRecords({ rm: "2025/26", limit: 20 }),
  client.searchDocuments({ from_date: today, limit: 30 })
]);

Error Handling Strategy

try {
  const data = await client.fetchCalendarEvents({ from: today, tom: today });
} catch (error) {
  if (error.message.includes('timeout')) {
    // Cold start or server overload - retry after 60s
    console.log("⏳ Server cold start detected, retrying...");
    await new Promise(resolve => setTimeout(resolve, 60000));
    const data = await client.fetchCalendarEvents({ from: today, tom: today });
  } else if (error.message.includes('503')) {
    // Server maintenance - fall back to cached data
    console.log("⚠️ MCP server unavailable, using cached data");
    // Load from cache...
  } else {
    throw error;
  }
}

MCP Tool Categories (32 Total Tools):

Riksdag Tools (15 tools):
- get_ledamoter: Retrieve parliament member list with party affiliation
- get_ledamot_details: Detailed biography and service history
- search_ledamoter: Member search by name, party, constituency
- get_motioner: All parliamentary motions with sponsorship
- search_motioner: Full-text motion search
- get_propositioner: Government legislative proposals
- search_propositioner: Proposal search by type, status
- get_dokument: Specific document retrieval
- search_dokument: Document discovery with metadata
- search_dokument_fulltext: Full-text document search
- get_voteringar: Voting records and roll-call results
- search_voteringar: Vote analysis by party, member, timeframe
- get_anforanden: Parliamentary speeches and debates
- search_anforanden: Speech search by speaker, topic, date
- get_calendar_events: Parliamentary calendar and scheduling

Government Tools (7 tools):
- get_regering_document: Retrieve government documents
- search_regering: Government document search
- search_regering_by_department: Department-specific searches
- summarize_regering_document: Automated government document summarization
- get_g0v_document_content: Markdown conversion of government documents
- get_g0v_document_types: List available document type categories
- analyze_g0v_by_department: Department-level document analysis

Statistical & Metadata Tools (5 tools):
- get_utskott: Committee list and composition
- get_betankanden: Committee reports and decisions
- fetch_report: Statistical reports (demographics, activity metrics)
- get_voting_group: Voting pattern analysis by party/constituency
- get_sync_status: Data freshness and update schedule status

Utility Tools (5 tools):
- get_data_dictionary: Schema definitions for all data types
- get_latest_update: Last successful data synchronization
- list_reports: Available statistical report types
- batch_fetch_documents: Efficient bulk document retrieval
- fetch_paginated_documents: Pagination-based result streaming

Protocol Details:

Direct Server Mode (Recommended):
- POST to https://riksdag-regering-ai.onrender.com/mcp
- Use unprefixed tool names (e.g., 'get_calendar_events')
- JSON-RPC 2.0 request format
- Automatic timeout handling and retries
- Lower latency (~200-500ms per request)

MCP Gateway Mode (Agentic Workflows):
- For agentic workflow sandbox environments with firewall container
- POST to http://host.docker.internal:80/mcp/riksdag-regering
- Use prefixed tool names: 'riksdag-regering--{tool_name}'
- Additional proxy latency (~50-200ms overhead per request)
- Auto-detection based on URL (checks for 'host.docker.internal')
- Client automatically handles prefixing - no manual changes needed

Architecture Comparison:

Direct Mode:
  Agent → HTTPS → MCP Server (riksdag-regering-ai.onrender.com)
  - Pros: Lower latency, simpler auth, faster cold start recovery
  - Cons: Less network control, requires explicit domain allowlist

Gateway Mode:
  Agent → Firewall → HTTP → Proxy → HTTPS → MCP Server
  - Pros: Security filtering, network audit, rate limiting
  - Cons: Higher latency, complex session mgmt, gateway timeout cascade

Auto-detection based on error responses

Intelligence Application Examples:

Automated News Generation:
- get_calendar_events → Week-ahead parliamentary schedule
- search_dokument → Recent propositions and motions
- search_anforanden → Parliamentary debate summaries
- get_voteringar → Party voting patterns and consensus analysis

Statistical Intelligence:
- get_ledamoter → Member demographics (age, gender, party)
- get_utskott → Committee composition and specialization
- fetch_report → Aggregated voting and productivity statistics
- get_voting_group → Coalition formation and party dynamics

Government Transparency:
- search_regering → Government policy announcements
- get_regering_document → Complete policy documentation
- summarize_regering_document → Automated policy summarization
- analyze_g0v_by_department → Department-specific tracking

Advanced Analysis:
- search_dokument_fulltext → Cross-document pattern matching
- search_voteringar → Historical voting pattern analysis
- get_anforanden → Debate transcript analysis
- batch_fetch_documents → Bulk data collection for trend analysis

Error Handling Strategy:
- Network errors: Automatic retries with exponential backoff (max 3 attempts)
- Timeout: 30-second default (adjustable via MCP_CLIENT_TIMEOUT_MS)
- Tool not found: Fallback from prefixed to non-prefixed names
- Rate limits: Respect server 429 responses with adaptive delays
- Server unavailable: Return cached data if available, else error