Improve system prompt to find redaction suggestions

Author: Peter Kumberger

src/lib/chat-tools.ts

@@ -148,8 +148,8 @@ export function executeAskUser(args: Record<string, unknown>): { special: Specia
   }
 }
 
-export function executeSuggestRedactions(args: Record<string, unknown>): { special: SpecialToolResult; toolResult: ToolResult } {
-  const raw = args.suggestions as Array<Record<string, unknown>>
+export function executeSuggestRedactions(args: Record<string, unknown>): { special: SpecialToolResult | null; toolResult: ToolResult } {
+  const raw = (args.suggestions as Array<Record<string, unknown>> | undefined) ?? []
   const suggestions: RedactionSuggestion[] = raw.map(s => ({
     documentKey: s.documentKey as string | undefined,
     text: s.text as string,
@@ -184,6 +184,18 @@ export function executeSuggestRedactions(args: Record<string, unknown>): { speci
   }))
   const remove = (args.remove as string[] | undefined) ?? []
   const total = suggestions.length + textRanges.length + pageRanges.length
+  // An empty call is always a protocol error: a document with nothing to redact
+  // is reported in prose instead of through this tool, so reaching here means
+  // the arguments failed to serialise rather than that the document is clean.
+  if (total === 0 && remove.length === 0) {
+    return {
+      special: null,
+      toolResult: {
+        success: false,
+        error: 'This call carried no arguments, so nothing was applied. If the document genuinely contains nothing to redact, do not call this tool at all — say so in your reply instead. If you did intend to suggest redactions, repeat the call with "suggestions" (or "textRanges"/"pageRanges") populated; each entry needs documentKey, the exact text copied from the read_documents response, pageIndex, confidence, person and personGroup.',
+      },
+    }
+  }
   return {
     special: { type: 'suggest_redactions', suggestions, textRanges, pageRanges, remove },
     toolResult: { success: true, data: `${total} Vorschläge hinzugefügt (${suggestions.length} Textstellen, ${textRanges.length} Textbereiche, ${pageRanges.length} Seitenbereiche), ${remove.length} entfernt.` },

src/lib/system-prompt.ts

@@ -18,13 +18,20 @@ export function buildSystemPrompt(opts: {
 }): string {
   const { redactionMode, foiJurisdiction, foiRules, locale } = opts
 
+  // `reason` is often just a restatement of `title` (e.g. "Personenbezogene Daten"),
+  // so the statutory text carries the criteria the model actually needs to apply.
+  const formatRule = (r: RedactionRule) => {
+    const summary = r.reason && r.reason !== r.title ? r.reason : undefined
+    const heading = `- **${r.title}**${r.reference ? ` (${r.reference})` : ''}${summary ? `: ${summary}` : ''}`
+    const detail = r.full_text?.replace(/\s*\n+\s*/g, ' ').trim()
+    return detail ? `${heading}\n  ${detail}` : heading
+  }
+
   const foiSection = redactionMode === 'foi'
     ? [
         '## FOI Mode',
         `Legal basis: ${foiJurisdiction ?? 'not selected'}`,
-        foiRules?.length
-          ? foiRules.map(r => `- **${r.title}** (${r.reference ?? ''}): ${r.reason ?? r.full_text ?? ''}`).join('\n')
-          : '',
+        foiRules?.length ? foiRules.map(formatRule).join('\n') : '',
       ].join('\n')
     : ''
 
@@ -84,6 +91,8 @@ export function buildSystemPrompt(opts: {
     '',
     'Use "low" ONLY for genuinely ambiguous individual cases in the document, not as a blanket rating.',
     '',
+    'If the document genuinely contains nothing that needs redacting, do NOT call `suggest_redactions` with empty arguments — say so in your reply instead.',
+    '',
     '## Tool discipline',
     '',
     '- Execute **only one tool call** per response.',

src/middleware.ts

@@ -4,9 +4,7 @@ import { routing } from './i18n/routing'
 export default createMiddleware(routing)
 
 export const config = {
-  // The `matcher` is relative to the `basePath`. The explicit '/' entry is
-  // required so the middleware fires on the base-path root (e.g. `/easyredact/`);
-  // Next.js does not trigger middleware on the root when a `basePath` is set and
-  // the matcher uses a negative-lookahead regex. See vercel/next.js#50161.
+  // Next.js skips the middleware on the base-path root when the matcher is a
+  // negative-lookahead regex, so '/' has to be listed explicitly.
   matcher: ['/', '/((?!api|_next|_vercel|.*\\..*).*)'],
 }