fix(#27): address security issues in intent classification

- Add input sanitization to prevent LLM prompt injection
  (escapes quotes, backslashes, replaces newlines)
- Add MaxLength(500) validation to DTO to prevent DoS
- Add entity validation to filter malicious LLM responses
- Add confidence validation to clamp values to 0.0-1.0
- Make LLM model configurable via INTENT_CLASSIFICATION_MODEL env var
- Add 12 new security tests (total: 72 tests, from 60)

Security fixes identified by code review:
- CVE-mitigated: Prompt injection via unescaped user input
- CVE-mitigated: Unvalidated entity data from LLM response
- CVE-mitigated: Missing input length validation

Co-Authored-By: Claude Opus 4.5 <[email protected]>

apps/api/src/brain/dto/intent-classification.dto.ts

@@ -1,12 +1,18 @@
-import { IsString, MinLength, IsOptional, IsBoolean } from "class-validator";
+import { IsString, MinLength, MaxLength, IsOptional, IsBoolean } from "class-validator";
 import type { IntentType, ExtractedEntity } from "../interfaces";
 
+/** Maximum query length to prevent DoS and excessive LLM costs */
+export const MAX_QUERY_LENGTH = 500;
+
 /**
  * DTO for intent classification request
  */
 export class ClassifyIntentDto {
   @IsString()
   @MinLength(1, { message: "query must not be empty" })
+  @MaxLength(MAX_QUERY_LENGTH, {
+    message: `query must not exceed ${String(MAX_QUERY_LENGTH)} characters`,
+  })
   query!: string;
 
   @IsOptional()