api-pentesting

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill repeatedly shows and requires full HTTP commands with Authorization/API-key placeholders (e.g., -H "Authorization: Bearer TOKEN", X-API-Key, REFRESH_TOKEN) and explicitly requires producing "full curl command to reproduce," which implies embedding real secret values verbatim in outputs.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). There are no direct download links (.exe/.msi/.dmg/etc.), but multiple attacker-controlled redirect/callback hosts (evil.com and app.com subdomain/redirect manipulations) are present and are suspicious because they can be used to host malware, steal OAuth tokens, or perform other malicious redirections.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This document is an offensive API-penetration playbook that explicitly instructs evading defenses, abusing auth flows, exfiltrating credentials (e.g., SSRF to cloud metadata, JWT jku/x5u attacker URLs, redirect_uri to attacker domains), performing mass-assignment privilege escalation, and to proceed without further permission — behaviors that enable deliberate malicious activity and credential/data theft.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (low risk: 0.10). The required runtime workflow ingests outsider-authored free text only via the target’s responses it actively fetches for testing (e.g., API/doc/schema discovery and GraphQL introspection/error messages), but it does not define any inbox/feed/browser/search intake where an outsider can post content for ingestion “without selection.”