p24

src/Agent/AgentRunner.php

@@ -8,6 +8,7 @@ use App\Commerce\Dto\ShopProductResult;
 use App\Commerce\SearchRepairService;
 use App\Commerce\ShopSearchService;
 use App\Config\AgentRunnerConfig;
+use App\Config\LanguageCleanupConfig;
 use App\Context\ContextService;
 use App\Context\UrlAnalyzer;
 use App\Infrastructure\OllamaClient;
@@ -33,6 +34,7 @@ final readonly class AgentRunner
         private OllamaClient $ollamaClient,
         private LoggerInterface $agentLogger,
         private AgentRunnerConfig $agentRunnerConfig,
+        private LanguageCleanupConfig $languageCleanupConfig,
         private bool $debug,
         private bool $logPrompt,
         private bool $logContext,
@@ -2710,7 +2712,7 @@ final readonly class AgentRunner
         $normalizedPrompt = $this->normalizeRagEvidenceText($prompt);
         $stopTerms = [];
 
-        foreach ($this->agentRunnerConfig->getRagEvidenceStopTerms() as $term) {
+        foreach ($this->getRagEvidenceCleanupStopTerms() as $term) {
             $term = $this->normalizeRagEvidenceText($term);
             if ($term !== '') {
                 $stopTerms[$term] = true;
@@ -2743,6 +2745,35 @@ final readonly class AgentRunner
         return $groups;
     }
 
+    /** @return string[] */
+    private function getRagEvidenceCleanupStopTerms(): array
+    {
+        return $this->mergeUniqueStrings(
+            $this->languageCleanupConfig->getStopWordsForProfile($this->agentRunnerConfig->getRagEvidenceCleanupProfile()),
+            $this->agentRunnerConfig->getRagEvidenceStopTerms()
+        );
+    }
+
+    /**
+     * @param string[] $left
+     * @param string[] $right
+     * @return string[]
+     */
+    private function mergeUniqueStrings(array $left, array $right): array
+    {
+        $out = [];
+        foreach (array_merge($left, $right) as $item) {
+            $item = trim((string) $item);
+            if ($item === '' || isset($out[$item])) {
+                continue;
+            }
+
+            $out[$item] = $item;
+        }
+
+        return array_values($out);
+    }
+
     /**
      * @return array<string, string[]>
      */

src/Config/AgentRunnerConfig.php

@@ -503,6 +503,11 @@ final class AgentRunnerConfig
         return $this->getRequiredInt('no_llm_fallback.max_shop_results');
     }
 
+    public function getRagEvidenceCleanupProfile(): string
+    {
+        return $this->getRequiredString('rag_evidence_guard.cleanup_profile');
+    }
+
     /**
      * @return string[]
      */

src/Config/RetriexEffectiveConfigProvider.php

@@ -496,6 +496,7 @@ final readonly class RetriexEffectiveConfigProvider
                 'debug_internal_error_prefix' => $this->agentRunnerConfig->getDebugInternalErrorPrefix(),
             ],
             'rag_evidence_guard' => [
+                'cleanup_profile' => $this->agentRunnerConfig->getRagEvidenceCleanupProfile(),
                 'stop_terms' => $this->agentRunnerConfig->getRagEvidenceStopTerms(),
                 'synonyms' => $this->agentRunnerConfig->getRagEvidenceSynonyms(),
                 'aggregate_query_patterns' => $this->agentRunnerConfig->getRagEvidenceAggregateQueryPatterns(),
@@ -1076,6 +1077,13 @@ final readonly class RetriexEffectiveConfigProvider
         }
 
         $ragEvidence = is_array($agent['rag_evidence_guard'] ?? null) ? $agent['rag_evidence_guard'] : [];
+        $ragEvidenceCleanupProfile = $ragEvidence['cleanup_profile'] ?? null;
+        if (!is_string($ragEvidenceCleanupProfile) || trim($ragEvidenceCleanupProfile) === '') {
+            $errors[] = 'agent.rag_evidence_guard.cleanup_profile must be a non-empty string.';
+        } elseif (!in_array($ragEvidenceCleanupProfile, $this->languageCleanupConfig->getCleanupProfileNames(), true)) {
+            $errors[] = 'agent.rag_evidence_guard.cleanup_profile references unknown language cleanup profile: ' . $ragEvidenceCleanupProfile . '.';
+        }
+
         $this->validateStringList($this->toList($ragEvidence['stop_terms'] ?? []), 'agent.rag_evidence_guard.stop_terms', $errors, $warnings);
         $this->validateStringListMap($ragEvidence['synonyms'] ?? [], 'agent.rag_evidence_guard.synonyms', $errors, $warnings);
         $this->validateRegexPatternList($ragEvidence['aggregate_query_patterns'] ?? [], 'agent.rag_evidence_guard.aggregate_query_patterns', $errors);