exploiting-race-condition-vulnerabilities

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill includes and demonstrates embedding session cookies/credentials directly in HTTP requests and example scripts (e.g., Cookie: session=VALID_SESSION and SESSION_COOKIE = "session=abc123"), which requires the LLM to place secret values verbatim into generated requests or code, creating exfiltration risk.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The repository contains explicit, actionable exploit code and examples for abusing race conditions to commit fraud and account takeover (coupon double-spend, overdraft, MFA bypass, email/password-reset race), indicating high potential for deliberate malicious use.

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

• Third-party content exposure detected (high risk: 0.85). The required workflow in scripts/agent.py sends requests to a user-supplied --url and then ingests the target’s returned resp.text[:200] (outsider-authored HTTP response free text) into the agent’s runtime data structures, which would be available for any LLM context building.