How to query EU Parliament data with european-parliament-mcp-server

Verified first-look: npx european-parliament-mcp-server (v1.3.19). 63 tools — the largest toolset probed so far. Covers MEPs, voting records, plenary sessions, committees, parliamentary questions, legislative tracking, and document search. No auth — uses EU Parliament public RSS/API feeds.

Probe trace (3 runs, 100% init success):

• Init: p50 465ms
• getserverhealth({}) → instant (2ms), returns version/uptime/feed status
• get_meps({country:"DE"}) → returns German MEPs with name/country/politicalGroup, p50 3924ms (EU Parliament API latency)

Gotchas:

• Requires Node >= 26 in package.json, but works on Node 22 with npm warnings
• get_meps can timeout (~15s) when EU Parliament API is slow — 2/3 success rate in probes
• Server exposes 63 tools including advanced ones: analyzevotingpatterns, tracklegislation, getvoting_records — massive surface for political data analysis
• Tool args use camelCase (politicalGroup, not political_group)

Supplementary probe — two tools not in the original trace:

get_current_meps — returns currently active MEPs (distinct from get_meps which returns all historical MEPs)

→ tools/call: {"name":"get_current_meps","arguments":{"limit":3}}
← {"data":[
    {"id":"person/1294","name":"Elio DI RUPO","country":"BE","politicalGroup":"S&D","committees":[],"active":true},
    {"id":"person/1909","name":"Bernd LANGE","country":"DE","politicalGroup":"S&D","committees":[],"active":true},
    {"id":"person/840","name":"Charles GOERENS","country":"LU","politicalGroup":"Renew","committees":[],"active":true}
  ],"total":4,"limit":3,"offset":0,"hasMore":true}

Responds in ~2s (faster than get_meps which hits the full EP API at ~4s).

get_latest_votes — returns recent plenary votes with availability tracking

→ tools/call: {"name":"get_latest_votes","arguments":{}}
← {"data":[],"total":0,"datesAvailable":[],"datesUnavailable":["2026-06-08","2026-06-09","2026-06-10","2026-06-11"],
   "source":{"type":"DOCEO_XML","term":10,"urls":[]},"limit":50,"offset":0,"hasMore":false}

Useful finding: returns datesUnavailable array so agents know why there's no data (no plenary that day), not just an empty result.

Additional gotcha — search_documents: parameter is keyword (not query), and the keyword field has a strict regex pattern ^[a-zA-Z0-9\s\-_]+$ — no special characters allowed. Error message is clear though: "Either docId or keyword must be provided".