High-severity findings.

Audited release

1.0.8

Findings

2

Worst severity

high

Content hash

a2378734e87a59d3772e293c… Copy

Findings

• high Unauthenticated POST /bft/v1/analyze burns configured AI API credits

  missing_permission_callback

  AiChat registers POST bft/v1/analyze with permission_callback __return_true and immediately calls ClientFactory::createClient()->analyzeBazi($data) using site-configured API credentials. Unauthenticated callers can exhaust paid AI quota / incur cost (DoS) and obtain analysis output. No rate limit or nonce observed on this route.

  inc/classes/API/AiChat.php:40-48

  Recommendation

  Require a nonce/capability or signed customer session; add rate limiting; never expose paid API proxy without auth.

• medium Avatar status endpoint is public and keyed only by request_id

  information_disclosure

  GET /bft/v1/check-avatar/{request_id} uses __return_true and reads wp_bazi_avatar_result_* / wp_bazi_avatar_request_* options. Predictable/leaked request IDs disclose avatar generation status/URLs.

  inc/classes/API/AiChat.php:51-98

  Recommendation

  Use unguessable tokens and/or auth; avoid exposing image URLs broadly.

WP Registry hashes the installable build and reports on that exact bytes-for-bytes copy. Embargoed findings are withheld until they are disclosed, so a clean verdict means nothing public is outstanding. WP Manifest does not audit code itself.