Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 | 4x 34x 34x 32x 32x 32x 30x 30x 30x 30x 30x 30x 30x 30x 36x 36x 36x 130x 96x 66x 36x 36x 130x 36x 130x 36x 36x 30x 2x 21x 21x 21x 21x 21x 21x 77x 77x 77x 77x 77x 21x 21x 21x 17x 17x 130x 130x 130x 61x 61x 130x 130x 130x 17x 36x 36x 36x 130x 130x 36x 17x 17x 36x 36x 36x 36x 36x 36x 36x 130x 130x 130x 130x 17x 36x 36x 130x 130x 130x 130x 36x 130x 36x 61x 69x 130x 17x 17x 36x 21x 21x 21x 21x 17x 17x 17x 17x 130x 130x 130x 61x 17x 17x 17x 36x 36x 36x 36x 36x 36x 36x 36x 17x 17x 17x 17x 17x 17x 19x 17x 15x 15x 1x 2x 1x 1x 23x 23x 23x 23x 23x 53x 53x 53x 53x 15x 7x 8x 38x 38x 23x 4x 23x 23x 3x 3x 3x 20x 1x 19x 2x 2x 17x 17x 24x 24x 24x 54x 23x 23x 6x 17x 17x 17x 17x 17x 36x 17x 24x 24x 2x 17x 1x 17x 17x 17x 1x 1x 4x 29x 29x | /**
* MCP Tool: comparative_intelligence
*
* Cross-reference MEP activities across committees, delegations, and
* legislative procedures for comprehensive multi-dimensional profiling.
* Enables side-by-side comparison and cluster analysis of 2-10 MEPs.
*
* **Intelligence Perspective:** Comparative intelligence enables ranking,
* outlier detection, and natural clustering of MEPs—essential for
* identifying coalition partners, influence leaders, and behavioral outliers.
*
* **Business Perspective:** Valuable for B2G/B2B/B2C clients—including
* government affairs teams, lobbying firms, and strategic consultancies—
* who need objective, data-driven MEP benchmarking for stakeholder
* prioritisation, engagement planning, and political risk scoring.
*
* **Marketing Perspective:** Compelling showcase for journalists, academic
* researchers, and civic-tech developers who want side-by-side MEP
* comparisons and cluster analysis without building custom analytics pipelines.
*
* ISMS Policy: SC-002 (Input Validation), AC-003 (Least Privilege)
*/
import { z } from 'zod';
import { epClient } from '../clients/europeanParliamentClient.js';
import { buildToolResponse, buildErrorResponse } from './shared/responseBuilder.js';
import type { ToolResult } from './shared/types.js';
import { ToolError } from './shared/errors.js';
import { handleToolError } from './shared/errorHandler.js';
import { extractHttpStatus } from './shared/errorClassifier.js';
export const ComparativeIntelligenceSchema = z.object({
mepIds: z.array(z.number().positive())
.min(2)
.max(10)
.describe('List of MEP IDs to compare (2-10 MEPs)'),
dimensions: z.array(
z.enum(['voting', 'committee', 'legislative', 'attendance'])
)
.optional()
.default(['voting', 'committee', 'legislative', 'attendance'])
.describe('Dimensions to include in the comparison')
});
export type ComparativeIntelligenceParams = z.infer<typeof ComparativeIntelligenceSchema>;
type Dimension = 'voting' | 'committee' | 'legislative' | 'attendance';
interface MepProfile {
mepId: string;
name: string;
politicalGroup: string;
country: string;
scores: Partial<Record<Dimension, number>>;
overallScore: number;
clusterLabel: string;
}
interface DimensionRanking {
dimension: Dimension;
ranking: { rank: number; mepId: string; name: string; score: number }[];
}
interface MepCluster {
clusterId: string;
members: string[];
characteristicDimension: Dimension;
avgScore: number;
}
interface PairScore {
mepA: string;
mepB: string;
similarityScore: number;
}
interface ComparativeIntelligenceResult {
mepCount: number;
dimensions: Dimension[];
profiles: MepProfile[];
rankingByDimension: DimensionRanking[];
correlationMatrix: PairScore[];
outlierMEPs: { mepId: string; name: string; outlierDimension: Dimension; outlierScore: number; zScore: number }[];
clusterAnalysis: MepCluster[];
computedAttributes: {
mostSimilarPair: { mepA: string; mepB: string; similarity: number | null };
mostDifferentPair: { mepA: string; mepB: string; similarity: number | null };
topOverallPerformer: string;
lowestOverallPerformer: string;
dimensionWithHighestVariance: Dimension;
};
dataAvailable: boolean;
confidenceLevel: 'HIGH' | 'MEDIUM' | 'LOW';
dataFreshness: string;
sourceAttribution: string;
methodology: string;
dataQualityWarnings: string[];
}
interface MEPApiData {
id: string;
name: string;
country: string;
politicalGroup: string;
committees: string[];
roles?: string[];
votingStatistics?: {
totalVotes: number;
votesFor: number;
abstentions: number;
attendanceRate: number;
};
};
/**
* Computes a normalized voting score (0–100) for an MEP based on
* participation volume and for-vote ratio.
*
* Scoring formula (equal weights):
* - **Participation** (50%) — capped linear scale: `min(100, totalVotes / 1500 × 100)`.
* Threshold 1500 is the approximate maximum votes cast in a full EP term, so an
* MEP attending every vote scores 100.
* - **For-vote ratio** (50%) — proportion of recorded votes cast *for* (`votesFor / totalVotes`).
* This is a simple yes-vote share and does **not** compare against party, group, or
* plenary majorities; it should not be interpreted as a formal party-line cohesion index.
*
* Current data limitation:
* - `epClient.getMEPDetails()` populates `votingStatistics` with placeholder values
* (all zeros) because the `/meps/{id}` endpoint does not expose real voting statistics.
* - Consequently, `stats.totalVotes` will typically be `0`, and this function will
* return `0` for most MEPs. Consumers MUST NOT treat this score as a meaningful
* voting metric until real EP voting data is integrated into the client/tooling.
*
* Returns `0` when no voting statistics are available, or when placeholder statistics
* with `totalVotes === 0` are supplied, to avoid misleading scores.
*
* @param mep - MEP API data record containing optional `votingStatistics`
* @returns Normalized score in the range `[0, 100]`, rounded to 2 decimal places;
* typically `0` due to EP API data limitations
*
* @security Handles missing or zeroed `votingStatistics` gracefully; never divides by zero
*/
function computeVotingScore(mep: MEPApiData): number {
const stats = mep.votingStatistics;
if (stats === undefined || stats.totalVotes === 0) return 0;
const participationScore = Math.min(100, (stats.totalVotes / 1500) * 100);
const forVoteRatioScore = (stats.votesFor / stats.totalVotes) * 100;
return Math.round((participationScore * 0.5 + forVoteRatioScore * 0.5) * 100) / 100;
}
/**
* Computes a normalized committee engagement score (0–100) for an MEP.
*
* Scoring formula:
* - **Membership breadth** (60%) — `min(100, committeeCount × 20)`. Five committees → 100.
* - **Leadership roles** (40%) — `min(100, leadershipRoles × 25)`. Four chair/vice-chair
* roles → 100. Roles are detected by the keywords `'chair'` or `'vice'`.
*
* @param mep - MEP API data record containing `committees` and optional `roles`
* @returns Normalized score in the range `[0, 100]`, rounded to 2 decimal places
*/
function computeCommitteeScore(mep: MEPApiData): number {
const membershipScore = Math.min(100, mep.committees.length * 20);
const roles = mep.roles ?? [];
const leadershipCount = roles.filter(r => r.toLowerCase().includes('chair') || r.toLowerCase().includes('vice')).length;
const leadershipScore = Math.min(100, leadershipCount * 25);
return Math.round((membershipScore * 0.6 + leadershipScore * 0.4) * 100) / 100;
}
/**
* Computes a normalized legislative output score (0–100) for an MEP.
*
* Uses a capped, point-based additive score (max 100):
* - **Rapporteurships** — 15 points each (6+ → 90 pts from this factor alone).
* - **Committee memberships** — 10 points each (broader presence amplifies legislative reach).
*
* Rapporteurships are detected by the `'rapporteur'` keyword in the MEP's roles list.
*
* @param mep - MEP API data record containing `committees` and optional `roles`
* @returns Normalized score in the range `[0, 100]`, rounded to 2 decimal places
*/
function computeLegislativeScore(mep: MEPApiData): number {
const roles = mep.roles ?? [];
const rapporteurships = roles.filter(r => r.toLowerCase().includes('rapporteur')).length;
return Math.round(Math.min(100, rapporteurships * 15 + mep.committees.length * 10) * 100) / 100;
}
/**
* Computes the attendance score for an MEP using `votingStatistics.attendanceRate` when available.
*
* The `attendanceRate` field is expected on a 0–100 scale, so this function applies
* rounding only—no additional scaling is performed.
*
* When `votingStatistics` or `attendanceRate` is absent, this function returns `0`.
* Because the current `/meps/{id}` EP endpoint does not expose attendance or voting
* statistics, callers should interpret a value of `0` as "attendance data currently
* unavailable" rather than a measured attendance rate.
*
* @param mep - MEP API data record containing optional `votingStatistics`
* @returns Numeric attendance score in the range `[0, 100]`; `0` typically indicates
* that attendance data is not available from the underlying EP API
*/
function computeAttendanceScore(mep: MEPApiData): number {
// attendanceRate is already in the 0-100 range; use it directly without scaling
return Math.round((mep.votingStatistics?.attendanceRate ?? 0) * 100) / 100;
}
/**
* Computes per-dimension scores for an MEP and returns them as a partial record.
*
* Delegates each dimension to the corresponding scoring function:
* - `'voting'` → {@link computeVotingScore}
* - `'committee'` → {@link computeCommitteeScore}
* - `'legislative'` → {@link computeLegislativeScore}
* - `'attendance'` → {@link computeAttendanceScore}
*
* @param mep - MEP API data record to score
* @param dimensions - Ordered list of dimensions to evaluate
* @returns Partial record mapping each requested dimension to its score (0–100)
*/
function computeDimensionScores(mep: MEPApiData, dimensions: Dimension[]): Partial<Record<Dimension, number>> {
const scores: Partial<Record<Dimension, number>> = {};
for (const dim of dimensions) {
if (dim === 'voting') scores.voting = computeVotingScore(mep);
else if (dim === 'committee') scores.committee = computeCommitteeScore(mep);
else if (dim === 'legislative') scores.legislative = computeLegislativeScore(mep);
else scores.attendance = computeAttendanceScore(mep);
}
return scores;
}
/**
* Computes the unweighted mean of all defined dimension scores.
*
* Undefined scores (e.g., a dimension that was not requested) are filtered out
* before averaging to avoid artificially penalising MEPs for missing dimensions.
*
* @param scores - Partial dimension score record from {@link computeDimensionScores}
* @returns Mean score in `[0, 100]` rounded to 2 decimal places, or `0` if empty
*/
function computeOverallScore(scores: Partial<Record<Dimension, number>>): number {
const values = (Object.values(scores) as (number | undefined)[]).filter((v): v is number => v !== undefined);
Iif (values.length === 0) return 0;
return Math.round((values.reduce((s, v) => s + v, 0) / values.length) * 100) / 100;
}
/**
* Assigns a human-readable performance-tier cluster label to an MEP.
*
* Labels follow the pattern `<tier>_<sanitised_group>` so downstream consumers
* can group MEPs by both performance and political affiliation:
* - **`high_performer_*`** — overall score ≥ 60
* - **`moderate_performer_*`** — overall score in [30, 60)
* - **`low_data_profile`** — score < 30 or insufficient data
*
* The political group name is lower-cased and non-letter characters are replaced
* with underscores before embedding in the label.
*
* @param overallScore - MEP's computed overall score (0–100)
* @param politicalGroup - Raw political group string (e.g., `'EPP'`, `'S&D'`)
* @returns Cluster label string suitable for display and grouping
*/
function assignClusterLabel(overallScore: number, politicalGroup: string): string {
const safeGroup = politicalGroup.toLowerCase().replace(/[^a-z]/g, '_');
if (overallScore >= 60) return `high_performer_${safeGroup}`;
if (overallScore >= 30) return `moderate_performer_${safeGroup}`;
return 'low_data_profile';
}
/**
* Computes cosine similarity between two MEP dimension-score vectors.
*
* Cosine similarity measures the angle between two vectors independent of
* magnitude — two MEPs with proportionally identical scores across all
* dimensions produce a similarity of 1.0 regardless of absolute score levels.
*
* Dimensions present in `scoresA` but absent from `scoresB` are treated as 0
* for `scoresB`. Dimensions absent from `scoresA` are ignored entirely.
*
* @param scoresA - Dimension scores for MEP A
* @param scoresB - Dimension scores for MEP B
* @returns Cosine similarity in `[0, 1]` rounded to 2 decimal places,
* or `0` if either vector is all-zero or empty
*/
function computeSimilarity(scoresA: Partial<Record<Dimension, number>>, scoresB: Partial<Record<Dimension, number>>): number {
const dims = Object.keys(scoresA) as Dimension[];
Iif (dims.length === 0) return 0;
let dotProduct = 0;
let normA = 0;
let normB = 0;
for (const dim of dims) {
const a = scoresA[dim] ?? 0;
const b = scoresB[dim] ?? 0;
dotProduct += a * b;
normA += a * a;
normB += b * b;
}
const denominator = Math.sqrt(normA) * Math.sqrt(normB);
Iif (denominator === 0) return 0;
return Math.round((dotProduct / denominator) * 100) / 100;
}
/**
* Identifies statistical outliers across all MEP profiles and dimensions.
*
* Uses **z-score outlier detection**: for each dimension, computes the population
* mean and standard deviation across all profiles, then flags any MEP whose score
* deviates by |z| ≥ 1.5 from the mean.
*
* **Threshold rationale:** A z-score threshold of 1.5 captures approximately 6.7%
* of a normal distribution in each tail (~13.4% two-sided) — a balanced sensitivity
* that surfaces meaningful outliers without flagging too many borderline cases.
* Stricter thresholds (2.0+) would miss genuine policy outliers in small groups.
*
* Results are sorted by |z-score| descending and capped at the top 5 outlier entries
* to keep the response concise.
*
* @param profiles - Array of computed MEP profiles (must contain at least 2 entries
* for meaningful statistics)
* @param dimensions - Dimensions to evaluate for outliers
* @returns Array of up to 5 outlier records, each containing the MEP identity,
* outlier dimension, raw score, and rounded z-score
*/
function detectOutliers(profiles: MepProfile[], dimensions: Dimension[]): ComparativeIntelligenceResult['outlierMEPs'] {
const outliers: ComparativeIntelligenceResult['outlierMEPs'] = [];
for (const dim of dimensions) {
const scores = profiles.map(p => p.scores[dim] ?? 0);
const mean = scores.reduce((s, v) => s + v, 0) / scores.length;
const variance = scores.reduce((s, v) => s + (v - mean) ** 2, 0) / scores.length;
const stdDev = Math.sqrt(variance);
for (const profile of profiles) {
const score = profile.scores[dim] ?? 0;
const zScore = stdDev > 0 ? (score - mean) / stdDev : 0;
Iif (Math.abs(zScore) >= 1.5) { // z ≥ 1.5σ: captures ~6.7% in each tail (~13.4% combined); balanced outlier sensitivity
outliers.push({ mepId: profile.mepId, name: profile.name, outlierDimension: dim, outlierScore: score, zScore: Math.round(zScore * 100) / 100 });
}
}
}
return outliers.sort((a, b) => Math.abs(b.zScore) - Math.abs(a.zScore)).slice(0, 5);
}
/**
* Returns the dimension with the highest average score across all cluster members.
*
* Used to characterise a cluster by its strongest dimension — e.g., a cluster
* dominated by committee chairs would surface `'committee'` as its characteristic.
*
* @param dimAvgs - Map of dimension → average score for the cluster
* @param dimensions - Full ordered list of dimensions (provides a safe fallback)
* @returns Dimension key with the highest average score, or the first dimension
* if all averages are ≤ 0
*/
function findBestDimension(dimAvgs: Partial<Record<Dimension, number>>, dimensions: Dimension[]): Dimension {
let best: Dimension = dimensions[0] ?? 'voting';
let bestAvg = -1;
for (const dim of dimensions) {
const avg = dimAvgs[dim] ?? 0;
if (avg > bestAvg) { bestAvg = avg; best = dim; }
}
return best;
}
/**
* Groups MEP profiles into clusters based on their pre-assigned `clusterLabel` and
* computes per-cluster statistics.
*
* Cluster assignment uses a two-factor scheme: performance tier (`high/moderate/low`)
* combined with political group affiliation (from {@link assignClusterLabel}). For each
* cluster the function calculates:
* - Per-dimension average scores
* - Overall average score across all dimensions
* - The characteristic dimension (highest average, via {@link findBestDimension})
*
* This approach is deterministic and requires no random seed, making it reproducible
* across runs with the same input.
*
* @param profiles - Array of scored MEP profiles (each must have a `clusterLabel`)
* @param dimensions - Dimensions to aggregate within each cluster
* @returns Array of {@link MepCluster} objects, one per distinct cluster label
*/
function buildClusterAnalysis(profiles: MepProfile[], dimensions: Dimension[]): MepCluster[] {
const clusterMap = new Map<string, { members: string[]; dimScores: Partial<Record<Dimension, number[]>> }>();
for (const profile of profiles) {
const label = profile.clusterLabel;
Eif (!clusterMap.has(label)) {
clusterMap.set(label, { members: [], dimScores: {} });
}
const entry = clusterMap.get(label);
Iif (entry === undefined) continue;
entry.members.push(profile.mepId);
for (const dim of dimensions) {
const score = profile.scores[dim] ?? 0;
const existing = entry.dimScores[dim];
if (existing === undefined) {
entry.dimScores[dim] = [score];
} else E{
existing.push(score);
}
}
}
return Array.from(clusterMap.entries()).map(([clusterId, data]) => {
const dimAvgs: Partial<Record<Dimension, number>> = {};
for (const dim of dimensions) {
const vals = data.dimScores[dim] ?? [];
dimAvgs[dim] = vals.length > 0
? Math.round((vals.reduce((s, v) => s + v, 0) / vals.length) * 100) / 100
: 0;
}
const allDimAvg = (Object.values(dimAvgs) as (number | undefined)[]).filter((v): v is number => v !== undefined);
const overallAvg = allDimAvg.length > 0
? Math.round((allDimAvg.reduce((s, v) => s + v, 0) / allDimAvg.length) * 100) / 100
: 0;
return { clusterId, members: data.members, characteristicDimension: findBestDimension(dimAvgs, dimensions), avgScore: overallAvg };
});
}
/**
* Builds a zero-value {@link ComparativeIntelligenceResult} for cases where
* fewer than 2 valid MEP profiles could be constructed.
*
* All analytical fields (rankings, correlations, outliers, clusters) are empty.
* `dataAvailable` is set to `false` and `confidenceLevel` to `'LOW'` to signal
* that no meaningful comparison could be performed.
*
* @param profiles - Partial list of profiles that were successfully built (may be empty)
* @param dimensions - Dimensions originally requested by the caller
* @returns A safe empty result that can be returned directly to the MCP client
*/
function buildEmptyResult(profiles: MepProfile[], dimensions: Dimension[]): ComparativeIntelligenceResult {
return {
mepCount: profiles.length,
dimensions,
profiles,
rankingByDimension: [],
correlationMatrix: [],
outlierMEPs: [],
clusterAnalysis: [],
computedAttributes: {
mostSimilarPair: { mepA: 'N/A', mepB: 'N/A', similarity: null },
mostDifferentPair: { mepA: 'N/A', mepB: 'N/A', similarity: null },
topOverallPerformer: 'N/A',
lowestOverallPerformer: 'N/A',
dimensionWithHighestVariance: dimensions[0] ?? 'voting'
},
dataAvailable: false,
confidenceLevel: 'LOW',
dataFreshness: 'Insufficient MEP data retrieved',
sourceAttribution: 'European Parliament Open Data Portal - data.europarl.europa.eu',
methodology: 'Comparative intelligence requires at least 2 valid MEP profiles.',
dataQualityWarnings: ['Insufficient MEP data — fewer than 2 valid profiles retrieved from EP API'],
};
}
/**
* Creates a zero-score placeholder profile for an MEP whose API call failed.
*
* Placeholder profiles allow the comparison to continue with the MEPs that
* were successfully retrieved, while still accounting for the requested ID in
* the output (so consumers can see which MEPs produced no data).
*
* @param mepId - Numeric MEP ID that failed to resolve
* @returns Placeholder {@link MepProfile} with zeroed scores and `'low_data_profile'` label
*/
function buildPlaceholderProfile(mepId: number): MepProfile {
return {
mepId: String(mepId),
name: `MEP-${String(mepId)} (not found)`,
politicalGroup: 'Unknown',
country: 'Unknown',
scores: {},
overallScore: 0,
clusterLabel: 'low_data_profile'
};
}
/**
* Produces per-dimension ranked lists of all MEPs in descending score order.
*
* Each {@link DimensionRanking} entry contains an ordered array where `rank: 1`
* is the highest scorer. Ties are resolved by the underlying sort algorithm
* (stable in V8/Node.js ≥ 11).
*
* @param profiles - Scored MEP profiles to rank
* @param dimensions - Dimensions for which rankings should be produced
* @returns Array of ranking objects, one per dimension
*/
function buildRankings(profiles: MepProfile[], dimensions: Dimension[]): DimensionRanking[] {
return dimensions.map(dim => ({
dimension: dim,
ranking: [...profiles]
.sort((a, b) => (b.scores[dim] ?? 0) - (a.scores[dim] ?? 0))
.map((p, idx) => ({ rank: idx + 1, mepId: p.mepId, name: p.name, score: p.scores[dim] ?? 0 }))
}));
}
/**
* Computes pairwise cosine-similarity scores for all unique MEP pairs.
*
* Only the upper triangle of the similarity matrix is computed (i < j) to
* avoid duplicating symmetric entries. The result is an unordered list of
* {@link PairScore} objects, one per unique pair.
*
* @param profiles - Scored MEP profiles; order determines pair enumeration
* @returns Array of pairwise similarity records (length = n*(n-1)/2)
*/
function buildCorrelationMatrix(profiles: MepProfile[]): PairScore[] {
const matrix: PairScore[] = [];
for (let i = 0; i < profiles.length; i++) {
for (let j = i + 1; j < profiles.length; j++) {
const a = profiles[i];
const b = profiles[j];
Iif (a === undefined || b === undefined) continue;
matrix.push({ mepA: a.mepId, mepB: b.mepId, similarityScore: computeSimilarity(a.scores, b.scores) });
}
}
return matrix;
}
/**
* Identifies the dimension with the highest score variance across all profiles.
*
* High variance in a dimension indicates strong differentiation among the MEPs
* being compared — useful for surfacing which axis is most discriminating.
*
* @param profiles - Scored MEP profiles
* @param dimensions - Dimensions to evaluate; must contain at least one entry
* @returns Dimension key with the greatest population variance, or the first
* dimension if all variances are equal (or zero)
*/
function findHighestVarianceDim(profiles: MepProfile[], dimensions: Dimension[]): Dimension {
let best: Dimension = dimensions[0] ?? 'voting';
let bestVariance = -1;
for (const dim of dimensions) {
const vals = profiles.map(p => p.scores[dim] ?? 0);
const mean = vals.reduce((s, v) => s + v, 0) / vals.length;
const variance = vals.reduce((s, v) => s + (v - mean) ** 2, 0) / vals.length;
if (variance > bestVariance) { bestVariance = variance; best = dim; }
}
return best;
}
/**
* Converts an array of settled `getMEPDetails` promises into scored MEP profiles.
*
* For each fulfilled result, dimension scores and an overall score are computed
* and a cluster label is assigned. For each rejected result (API error, unknown MEP),
* a zero-score placeholder profile is inserted to preserve the requested ID in the
* output.
*
* @param detailsResults - Settled promise results from `Promise.allSettled`
* @param mepIds - Original MEP IDs in the same order as `detailsResults`
* @param dimensions - Dimensions to score for each successfully resolved MEP
* @returns Array of {@link MepProfile} objects in the same order as `mepIds`
*/
function buildProfilesFromResults(
detailsResults: PromiseSettledResult<unknown>[],
mepIds: number[],
dimensions: Dimension[]
): MepProfile[] {
const profiles: MepProfile[] = [];
for (let i = 0; i < detailsResults.length; i++) {
const result = detailsResults[i];
const mepId = mepIds[i];
Iif (result === undefined || mepId === undefined) continue;
if (result.status === 'fulfilled') {
const mep = result.value as MEPApiData;
const scores = computeDimensionScores(mep, dimensions);
const overallScore = computeOverallScore(scores);
profiles.push({
mepId: mep.id,
name: mep.name,
politicalGroup: mep.politicalGroup,
country: mep.country,
scores,
overallScore,
clusterLabel: assignClusterLabel(overallScore, mep.politicalGroup)
});
} else E{
profiles.push(buildPlaceholderProfile(mepId));
}
}
return profiles;
}
/**
* Derives the `computedAttributes` block of the comparative intelligence result.
*
* Identifies:
* - **Most similar pair** — highest-scoring entry in the sorted correlation matrix
* - **Most different pair** — lowest-scoring entry in the sorted correlation matrix
* - **Top/lowest overall performers** — sorted by `overallScore` descending
* - **Dimension with highest variance** — via {@link findHighestVarianceDim}
*
* Defaults to `'N/A'` / 0 for any field that cannot be derived from an empty matrix.
*
* @param profiles - Scored MEP profiles
* @param correlationMatrix - Pre-computed pairwise similarity scores
* @param dimensions - Dimensions available for variance calculation
* @returns Derived summary attributes for the comparison result
*/
function buildComputedAttributes(
profiles: MepProfile[],
correlationMatrix: PairScore[],
dimensions: Dimension[]
): ComparativeIntelligenceResult['computedAttributes'] {
const sortedCorr = [...correlationMatrix].sort((a, b) => b.similarityScore - a.similarityScore);
const firstPair = sortedCorr[0];
const lastPair = sortedCorr[sortedCorr.length - 1];
const mostSimilarPair = firstPair !== undefined
? { mepA: firstPair.mepA, mepB: firstPair.mepB, similarity: firstPair.similarityScore }
: { mepA: 'N/A', mepB: 'N/A', similarity: null };
const mostDifferentPair = lastPair !== undefined
? { mepA: lastPair.mepA, mepB: lastPair.mepB, similarity: lastPair.similarityScore }
: { mepA: 'N/A', mepB: 'N/A', similarity: null };
const sortedByOverall = [...profiles].sort((a, b) => b.overallScore - a.overallScore);
return {
mostSimilarPair,
mostDifferentPair,
topOverallPerformer: sortedByOverall[0]?.name ?? 'N/A',
lowestOverallPerformer: sortedByOverall[sortedByOverall.length - 1]?.name ?? 'N/A',
dimensionWithHighestVariance: findHighestVarianceDim(profiles, dimensions)
};
}
/**
* Determines whether a rejected promise reason represents a 404 (MEP not found)
* as opposed to a transient error (timeout, 429, 5xx).
*/
function isNotFoundError(reason: unknown): boolean {
const status = extractHttpStatus(reason);
return status === 404;
}
/**
* Builds an error ToolResult for when all MEP IDs failed with 404.
*/
function buildAllNotFoundError(notFoundMepIds: number[]): ToolResult {
return buildErrorResponse(
new ToolError({
toolName: 'comparative_intelligence',
operation: 'validateMEPIds',
message: `None of the provided MEP IDs could be found: ${notFoundMepIds.join(', ')}. Please verify the MEP IDs and try again.`,
isRetryable: false,
errorCode: 'INVALID_PARAMS',
}),
'comparative_intelligence'
);
}
/**
* Builds an error ToolResult for when fewer than 2 valid MEPs remain after
* excluding IDs that could not be resolved. When transient errors contributed
* to the shortfall, the error is retryable (a retry may yield enough valid MEPs).
*/
function buildInsufficientMepsError(
validCount: number,
totalCount: number,
unresolvedIds: number[],
hasTransientErrors: boolean,
transientCause?: unknown
): ToolResult {
if (hasTransientErrors) {
return buildTransientError(
`Only ${String(validCount)} of ${String(totalCount)} MEP IDs could be resolved. Comparative analysis requires at least 2 valid MEPs. Unresolved IDs: ${unresolvedIds.join(', ')}`,
transientCause
);
}
return buildErrorResponse(
new ToolError({
toolName: 'comparative_intelligence',
operation: 'validateMEPIds',
message: `Only ${String(validCount)} of ${String(totalCount)} MEP IDs could be resolved. Comparative analysis requires at least 2 valid MEPs. Unresolved IDs: ${unresolvedIds.join(', ')}`,
isRetryable: false,
errorCode: 'INVALID_PARAMS',
}),
'comparative_intelligence'
);
}
interface MepResultPartition {
notFoundMepIds: number[];
validResults: PromiseSettledResult<unknown>[];
validMepIds: number[];
transientErrors: { mepId: number; reason: unknown }[];
}
/**
* Partitions `Promise.allSettled` results into valid, not-found (404), and
* transient error buckets based on the rejection reason's status code.
*/
function partitionMepResults(
detailsResults: PromiseSettledResult<unknown>[],
mepIds: number[]
): MepResultPartition {
const notFoundMepIds: number[] = [];
const validResults: PromiseSettledResult<unknown>[] = [];
const validMepIds: number[] = [];
const transientErrors: { mepId: number; reason: unknown }[] = [];
for (let i = 0; i < detailsResults.length; i++) {
const result = detailsResults[i];
const mepId = mepIds[i];
Iif (result === undefined || mepId === undefined) continue;
if (result.status === 'rejected') {
if (isNotFoundError(result.reason)) {
notFoundMepIds.push(mepId);
} else {
transientErrors.push({ mepId, reason: result.reason });
}
} else {
validResults.push(result);
validMepIds.push(mepId);
}
}
return { notFoundMepIds, validResults, validMepIds, transientErrors };
}
/**
* Builds a retryable error ToolResult for upstream/transient failures.
*/
function buildTransientError(message: string, cause: unknown): ToolResult {
return handleToolError(
new ToolError({
toolName: 'comparative_intelligence',
operation: 'fetchMEPProfiles',
message,
isRetryable: true,
cause: cause instanceof Error ? cause : undefined,
}),
'comparative_intelligence'
);
}
/**
* Validates partitioned MEP results and returns either an error ToolResult
* or the valid partition for continued analysis.
*
* - **404 rejections** are treated as invalid MEP IDs (non-retryable).
* - **Non-404 rejections** (timeout, 429, 503, etc.) are treated as transient
* failures and propagated via `handleToolError` so retryability is correct.
* - When 2+ valid MEPs exist, transient-failed IDs are returned so callers
* can surface warnings about silently excluded IDs.
*/
function validateMepResults(
detailsResults: PromiseSettledResult<unknown>[],
mepIds: number[]
): { error: ToolResult } | { notFoundMepIds: number[]; transientFailedMepIds: number[]; validResults: PromiseSettledResult<unknown>[]; validMepIds: number[] } {
const { notFoundMepIds, validResults, validMepIds, transientErrors } =
partitionMepResults(detailsResults, mepIds);
// No valid MEPs: determine error type based on failure categories
if (validMepIds.length === 0 && transientErrors.length > 0) {
const cause = transientErrors[0]?.reason;
const message = notFoundMepIds.length === 0
? 'Failed to retrieve any MEP profiles due to upstream errors'
: `Failed to retrieve MEP profiles. ${String(notFoundMepIds.length)} MEP ID(s) not found: ${notFoundMepIds.join(', ')}. ${String(transientErrors.length)} MEP ID(s) failed due to upstream errors and may be retried.`;
return { error: buildTransientError(message, cause) };
}
if (validMepIds.length === 0) {
return { error: buildAllNotFoundError(notFoundMepIds) };
}
if (validMepIds.length < 2) {
const unresolvedIds = [...notFoundMepIds, ...transientErrors.map(e => e.mepId)];
return { error: buildInsufficientMepsError(validMepIds.length, mepIds.length, unresolvedIds, transientErrors.length > 0, transientErrors[0]?.reason) };
}
const transientFailedMepIds = transientErrors.map(e => e.mepId);
return { notFoundMepIds, transientFailedMepIds, validResults, validMepIds };
}
export async function comparativeIntelligence(params: ComparativeIntelligenceParams): Promise<ToolResult> {
try {
const dimensions = params.dimensions as Dimension[];
const detailsResults = await Promise.allSettled(
params.mepIds.map(id => epClient.getMEPDetails(String(id)))
);
// Early validation: identify which MEP IDs resolved and which failed
const validation = validateMepResults(detailsResults, params.mepIds);
if ('error' in validation) {
return validation.error;
}
const { notFoundMepIds, transientFailedMepIds, validResults, validMepIds } = validation;
// Build profiles only from valid (fulfilled) results — invalid/transient IDs
// are intentionally excluded; buildProfilesFromResults will find only fulfilled
// entries in this pre-filtered subset (no placeholder profiles are generated).
const profiles = buildProfilesFromResults(validResults, validMepIds, dimensions);
Iif (profiles.length < 2) {
return buildToolResponse(buildEmptyResult(profiles, dimensions));
}
const rankingByDimension = buildRankings(profiles, dimensions);
const correlationMatrix = buildCorrelationMatrix(profiles);
const dataWithScores = profiles.filter(p => p.overallScore > 0).length;
const confidenceLevel: 'HIGH' | 'MEDIUM' | 'LOW' =
dataWithScores >= validMepIds.length * 0.8 ? 'MEDIUM' : 'LOW';
const dataQualityWarnings = [
'Voting and attendance scores are placeholder zeros — EP API does not expose per-MEP voting statistics',
'Correlation matrix computed from partially-zero score vectors — similarity values may not be meaningful',
];
if (notFoundMepIds.length > 0) {
dataQualityWarnings.push(
`${String(notFoundMepIds.length)} MEP ID(s) could not be found and were excluded: ${notFoundMepIds.join(', ')}`
);
}
if (transientFailedMepIds.length > 0) {
dataQualityWarnings.push(
`${String(transientFailedMepIds.length)} MEP ID(s) could not be retrieved due to upstream errors and were excluded: ${transientFailedMepIds.join(', ')}`
);
}
const result: ComparativeIntelligenceResult = {
mepCount: profiles.length,
dimensions,
profiles,
rankingByDimension,
correlationMatrix,
outlierMEPs: detectOutliers(profiles, dimensions),
clusterAnalysis: buildClusterAnalysis(profiles, dimensions),
computedAttributes: buildComputedAttributes(profiles, correlationMatrix, dimensions),
dataAvailable: profiles.some(p => p.overallScore > 0),
confidenceLevel,
dataFreshness: 'Real-time EP API data — MEP profiles from current EP Open Data',
sourceAttribution: 'European Parliament Open Data Portal - data.europarl.europa.eu',
methodology: 'Multi-dimensional MEP profiling and comparison. '
+ 'Voting score: participation volume (50%) + for-vote rate (50%). '
+ 'Committee score: membership breadth (60%) + leadership roles (40%). '
+ 'Legislative score: rapporteurships (15 pts each) + committee count (10 pts each). '
+ 'Attendance score: placeholder only — EP /meps/{id} does not expose attendance data; value is 0 until real data is available. '
+ 'Similarity: cosine similarity between score vectors. '
+ 'Outliers: z-score ≥ 1.5 standard deviations from group mean. '
+ 'Clusters: grouped by political group + performance tier. '
+ 'NOTE: Voting and attendance statistics are not available from the current EP API; all related scores are placeholder zeros. '
+ 'Data source: https://data.europarl.europa.eu/api/v2/meps',
dataQualityWarnings,
};
return buildToolResponse(result);
} catch (error: unknown) {
const toolError = error instanceof ToolError
? error
: new ToolError({
toolName: 'comparative_intelligence',
operation: 'fetchMEPProfiles',
message: 'Failed to retrieve MEP profiles for comparison',
isRetryable: true,
cause: error,
});
return handleToolError(toolError, 'comparative_intelligence');
}
}
export const comparativeIntelligenceToolMetadata = {
name: 'comparative_intelligence',
description: 'Cross-reference 2-10 MEP activities across voting, committee, legislative, and attendance dimensions. Returns ranked profiles, correlation matrix, outlier detection, and cluster analysis for comprehensive comparative intelligence.',
inputSchema: {
type: 'object' as const,
properties: {
mepIds: {
type: 'array',
items: { type: 'number' },
description: 'List of MEP IDs to compare (2-10 MEPs)',
minItems: 2,
maxItems: 10
},
dimensions: {
type: 'array',
items: { type: 'string', enum: ['voting', 'committee', 'legislative', 'attendance'] },
description: 'Dimensions to include in the comparison',
default: ['voting', 'committee', 'legislative', 'attendance']
}
},
required: ['mepIds']
}
};
export async function handleComparativeIntelligence(args: unknown): Promise<ToolResult> {
const params = ComparativeIntelligenceSchema.parse(args);
return comparativeIntelligence(params);
}
|