symmetric-cipher-attacks

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). This is an explicit offensive cryptanalysis playbook with ready-to-run exploit code and step‑by‑step techniques (padding oracle, CBC bit‑flipping, ECB cut‑and‑paste/byte‑at‑a‑time, stream key reuse, PRNG/LFSR/MT recovery, GCM nonce reuse, etc.) that enable unauthorized decryption, forgery of ciphertexts, key/state recovery, and privilege escalation—highly abuseable for data exfiltration and account takeover.

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

• Third-party content exposure detected (high risk: 1.00). The skill's attack workflows (e.g., the padding oracle and ECB byte‑at‑a‑time routines) explicitly require sending requests to target endpoints and interpreting their responses — see the oracle(ct) calls in the Python examples and the PadBuster example "padbuster http://target/decrypt?ct=" in SKILL.md — which means the agent will fetch and act on untrusted third‑party server content to drive its next actions.