CapaKraken/packages/api/src/router/assistant-tool-selection.ts

import { MUTATION_TOOLS } from "./assistant-tools.js";
import type { ToolDef } from "./assistant-tools/shared.js";

const MAX_OPENAI_TOOL_DEFINITIONS = 128;

const ALWAYS_INCLUDED_TOOL_NAMES = new Set([
  "get_current_user",
  "get_resource",
  "search_projects",
  "get_project",
  "list_allocations",
  "get_statistics",
  "navigate_to_page",
]);

const MUTATION_INTENT_KEYWORDS = [
  "create", "add", "new", "update", "change", "edit", "delete", "remove", "cancel", "approve", "reject",
  "anlegen", "erstellen", "neu", "aendern", "ändern", "bearbeiten", "loeschen", "löschen", "entfernen",
  "stornieren", "genehmigen", "ablehnen", "setzen",
];

const TOOL_SELECTION_HINTS = [
  {
    keywords: ["holiday", "holidays", "feiertag", "feiertage", "vacation", "vacations", "urlaub", "ferien", "abwesen"],
    nameFragments: ["holiday", "vacation", "entitlement"],
    exactTools: ["list_holidays_by_region", "get_resource_holidays", "get_my_timeline_holiday_overlays", "list_holiday_calendars", "get_holiday_calendar", "preview_resolved_holiday_calendar"],
  },
  {
    keywords: ["resource", "resources", "ressource", "ressourcen", "employee", "mitarbeiter", "person", "people", "team", "chapter", "skill", "skills"],
    nameFragments: ["resource", "skill", "role", "user", "staffing", "capacity"],
    exactTools: ["search_resources", "get_resource", "search_by_skill", "check_resource_availability", "get_staffing_suggestions", "find_capacity"],
  },
  {
    keywords: ["capacity", "availability", "available", "kapazitaet", "kapazität", "verfuegbar", "verfügbar", "auslastung", "chargeability", "sah", "lcr"],
    nameFragments: ["capacity", "availability", "chargeability", "staffing", "rate", "budget"],
    exactTools: ["check_resource_availability", "get_staffing_suggestions", "find_capacity", "get_chargeability", "find_best_project_resource", "resolve_rate"],
  },
  {
    keywords: ["project", "projects", "projekt", "projekte", "allocation", "allocations", "allokation", "allokationen", "assignment", "assignments", "demand", "demands", "timeline"],
    nameFragments: ["project", "allocation", "demand", "timeline", "assignment", "blueprint"],
    exactTools: ["search_projects", "get_project", "list_allocations", "list_demands", "get_my_timeline_entries_view", "get_my_timeline_holiday_overlays", "get_timeline_entries_view", "get_project_timeline_context"],
  },
  {
    keywords: ["dashboard", "widget", "widgets", "peak", "forecast", "insight", "insights", "anomaly", "anomalies", "report", "reports", "analyse", "analysis", "bericht"],
    nameFragments: ["dashboard", "statistics", "report", "insight", "anomal", "health", "forecast", "skill"],
    exactTools: ["get_statistics", "get_dashboard_detail", "detect_anomalies", "get_skill_gaps", "get_project_health", "get_budget_forecast", "get_insights_summary", "run_report"],
  },
  {
    keywords: ["estimate", "estimates", "angebot", "angebote", "budget", "budgets", "cost", "costs", "kosten", "rate", "rates", "preis", "preise"],
    nameFragments: ["estimate", "budget", "rate", "cost"],
    exactTools: ["get_budget_status", "list_rate_cards", "resolve_rate", "lookup_rate", "search_estimates", "get_estimate_detail"],
  },
  {
    keywords: ["notification", "notifications", "benachrichtigung", "benachrichtigungen", "task", "tasks", "aufgabe", "aufgaben", "reminder", "reminders", "broadcast"],
    nameFragments: ["notification", "task", "reminder", "broadcast"],
    exactTools: ["list_notifications", "get_unread_notification_count", "list_tasks", "get_task_counts", "list_reminders", "get_broadcast_detail"],
  },
  {
    keywords: ["country", "countries", "land", "laender", "länder", "city", "cities", "stadt", "staedte", "städte", "region", "regions", "state", "bundesland"],
    nameFragments: ["country", "metro_city", "holiday_calendar"],
    exactTools: ["list_countries", "get_country", "list_holidays_by_region", "list_holiday_calendars"],
  },
  {
    keywords: ["user", "users", "permission", "permissions", "rolle", "rollen", "admin", "system", "webhook", "import", "audit", "history", "rechte"],
    nameFragments: ["user", "permission", "role", "system", "webhook", "import", "audit", "history", "org_unit", "country"],
    exactTools: ["list_users", "get_effective_user_permissions", "list_audit_log_entries", "query_change_history", "get_system_settings", "list_webhooks"],
  },
];

const TOOL_SELECTION_STOP_WORDS = new Set([
  "the", "and", "for", "with", "from", "that", "this", "what", "when", "where", "who", "how",
  "und", "der", "die", "das", "ein", "eine", "einer", "einem", "einen", "mit", "von", "fuer", "für",
  "auf", "ist", "sind", "im", "in", "am", "an", "zu", "zum", "zur", "mir", "bitte", "can", "you",
  "mir", "alle", "all", "den", "dem", "des",
]);

export type AssistantChatMessage = { role: "user" | "assistant"; content: string };

export function normalizeAssistantText(input: string): string {
  return input
    .toLowerCase()
    .normalize("NFD")
    .replace(/\p{Diacritic}/gu, " ")
    .replace(/[^a-z0-9_]+/g, " ")
    .replace(/\s+/g, " ")
    .trim();
}

export function tokenizeAssistantIntent(input: string): string[] {
  return normalizeAssistantText(input)
    .split(" ")
    .map((token) => token.trim())
    .filter((token) => token.length >= 3 && !TOOL_SELECTION_STOP_WORDS.has(token));
}

export function selectAssistantToolsForRequest(
  availableTools: ToolDef[],
  messages: AssistantChatMessage[],
  pageContext?: string,
): ToolDef[] {
  if (availableTools.length <= MAX_OPENAI_TOOL_DEFINITIONS) {
    return availableTools;
  }

  const recentUserText = messages
    .filter((message) => message.role === "user")
    .slice(-4)
    .map((message) => message.content)
    .join(" ");
  const intentText = [recentUserText, pageContext ?? ""].filter(Boolean).join(" ");
  const normalizedIntent = normalizeAssistantText(intentText);
  const intentTokens = tokenizeAssistantIntent(intentText);
  const mutationIntent = MUTATION_INTENT_KEYWORDS.some((keyword) => normalizedIntent.includes(normalizeAssistantText(keyword)));

  const selectedHintTools = new Set<string>();
  for (const hint of TOOL_SELECTION_HINTS) {
    const matchedKeyword = hint.keywords.some((keyword) => normalizedIntent.includes(normalizeAssistantText(keyword)));
    if (!matchedKeyword) continue;
    for (const toolName of hint.exactTools) {
      selectedHintTools.add(toolName);
    }
  }

  return availableTools
    .map((tool, index) => {
      const name = tool.function.name;
      const normalizedName = normalizeAssistantText(name.replace(/_/g, " "));
      const normalizedDescription = normalizeAssistantText(tool.function.description);
      let score = 0;

      if (ALWAYS_INCLUDED_TOOL_NAMES.has(name)) score += 1000;
      if (selectedHintTools.has(name)) score += 400;

      for (const hint of TOOL_SELECTION_HINTS) {
        const matchedKeyword = hint.keywords.some((keyword) => normalizedIntent.includes(normalizeAssistantText(keyword)));
        if (!matchedKeyword) continue;
        if (hint.exactTools.includes(name)) score += 160;
        if (hint.nameFragments.some((fragment) => name.includes(fragment))) score += 120;
        if (hint.nameFragments.some((fragment) => normalizedDescription.includes(normalizeAssistantText(fragment)))) score += 40;
      }

      for (const token of intentTokens) {
        if (normalizedName.includes(token)) score += 45;
        if (normalizedDescription.includes(token)) score += 10;
      }

      if (name.startsWith("search_")) score += 18;
      if (name.startsWith("get_")) score += 12;
      if (name.startsWith("list_")) score += 10;

      if (MUTATION_TOOLS.has(name)) {
        score += mutationIntent ? 40 : -30;
      } else {
        score += 8;
      }

      return { tool, index, score };
    })
    .sort((left, right) => {
      if (right.score !== left.score) return right.score - left.score;
      return left.index - right.index;
    })
    .slice(0, MAX_OPENAI_TOOL_DEFINITIONS)
    .map((entry) => entry.tool);
}