Security & Permissions

Designed for user control

Voycerizer aims to make conversational publishing convenient without bypassing the approvals required to act on a user's accounts. Its architecture separates the LLM host's intent from the backend that validates, authorizes, and executes platform requests.

Official connections

Social-platform permissions are requested using official provider authorization flows. Users should never paste account passwords, OAuth refresh tokens, or API secrets into the conversation to enable publishing.

Explicit publication decisions

Our intended flow requires destination-specific validation and the appropriate user or provider authority before any publishing mutation. Some platforms require additional settings, consent, or commercial disclosures.

Boundaries that matter

Tokens and privileged service credentials belong on the server, never in browser JavaScript or LLM tool output.
Every authenticated user's schedules, accounts and media must be isolated from other users.
Queued work should be idempotent so retries do not silently duplicate already-completed posts.
Publishing receipts should distinguish submitted, processing, live, draft-ready and failed states.
Platform-specific restrictions and approvals remain enforceable even when a chat asks to bypass them.

Development and approval status

The service is still preparing for public launch. The controls described above are design requirements, not a claim of independent certification or a completed production security audit. Platform access depends on provider-specific approvals.

Questions or reports