PT-2026-35599

Publicado

2026-04-17

·

Atualizado

2026-04-17

Descrição por

CVSS 3.1
7.5
Alta
MétricasAV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

Summary

The text-to-speech generation endpoint (POST /api/v1/text-to-speech/generate) is whitelisted (no auth) and accepts a credentialId directly in the request body. When called without a chatflowId, the endpoint uses the provided credentialId to decrypt the stored credential (e.g., OpenAI or ElevenLabs API key) and generate speech.

Root Cause

typescript
// packages/server/src/controllers/text-to-speech/index.ts:58-64
} else {
  // Use TTS config from request body
  provider = bodyProvider
  credentialId = bodyCredentialId // ← attacker-controlled credential ID
  voice = bodyVoice
  model = bodyModel
}

Docker Validation

POST /api/v1/text-to-speech/generate with arbitrary credentialId in body: endpoint processes request, sends SSE tts start event, only fails when credential doesn't exist — proves code path runs without authentication.

Impact

  • Use victim's API keys (OpenAI, ElevenLabs, Azure, Google) without authorization
  • Burn API credits on the victim's account
  • Generate unlimited speech content at victim's expense
  • Combined with credential ID leak from Finding 2, this is trivially exploitable

Suggested Fix

Remove the TTS endpoint from WHITELIST URLS or validate that the credential belongs to the chatflow being used:
typescript
// Only allow credentialId when it matches the chatflow's TTS configuration
if (!chatflowId) {
  return res.status(401).json({ message: 'Authentication required' })
}

References

  • packages/server/src/controllers/text-to-speech/index.ts lines 10-162
  • packages/server/src/utils/constants.ts line 41 (whitelist entry)

Credits

Explorabilidade
Correção
Impactos
Fornecedores
Produtos afetados
Identificadores relacionados
GHSA-5FW2-MWHH-9947
Encontrou algum problema na descrição? Tem algo a acrescentar? Fique à vontade para nos escrever 👾