Back to skill

Security audit

Skill Quality Auditor

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AgentSkill quality reviewer that reads a user-provided skill package and reports a scorecard without installing code, executing scripts, persisting data, or sending data elsewhere.

Install only if you want an agent to inspect the contents of skill packages you point it at. Expect it to read the reviewed skill's files and produce a Chinese-formatted scorecard unless you override the output language or adapt the template.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (11)

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The report templates hard-code Chinese output labels and structure, which forces a language choice without user opt-in and can override the surrounding system or user language preference. In an agent setting, this creates instruction-conflict behavior, reduces reliability, and can be abused as a form of output steering that disrupts downstream consumers expecting another language or format.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
90% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 36)May include surrounding context.

md
grep -rn -E 'curl\s.*\|\s*(ba|da)sh|wget\s.*\|\s*(ba|da)sh|eval\s.*\$\(curl|pip\s+install\s+https?://' /path/to/skill/

# C4: Sensitive paths (path-format only, avoid false positives)
grep -rn -E '/\.ssh/|/\.aws/credentials|/etc/passwd|/etc/shadow' /path/to/skill/

# C5: Crypto addresses (flag for manual review — high false positive rate)
grep -rn -E '0x[a-fA-F0-9]{40}' /path/to/skill/

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 36)May include surrounding context.

md
grep -rn -E 'curl\s.*\|\s*(ba|da)sh|wget\s.*\|\s*(ba|da)sh|eval\s.*\$\(curl|pip\s+install\s+https?://' /path/to/skill/

# C4: Sensitive paths (path-format only, avoid false positives)
grep -rn -E '/\.ssh/|/\.aws/credentials|/etc/passwd|/etc/shadow' /path/to/skill/

# C5: Crypto addresses (flag for manual review — high false positive rate)
grep -rn -E '0x[a-fA-F0-9]{40}' /path/to/skill/

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 66)May include surrounding context.

md
grep -rn -E 'curl\s.*\|\s*(ba|da)sh|wget\s.*\|\s*(ba|da)sh|eval\s.*\$\(curl|pip\s+install\s+https?://' /path/to/skill/

# C4: Sensitive paths (path-format only, avoid false positives)
grep -rn -E '/\.ssh/|/\.aws/credentials|/etc/passwd|/etc/shadow' /path/to/skill/

# C5: Crypto addresses (flag for manual review — high false positive rate)
grep -rn -E '0x[a-fA-F0-9]{40}' /path/to/skill/

Credential Access

High
Category
Privilege Escalation
Confidence
95% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 36)May include surrounding context.

md
grep -rn -E 'curl\s.*\|\s*(ba|da)sh|wget\s.*\|\s*(ba|da)sh|eval\s.*\$\(curl|pip\s+install\s+https?://' /path/to/skill/

# C4: Sensitive paths (path-format only, avoid false positives)
grep -rn -E '/\.ssh/|/\.aws/credentials|/etc/passwd|/etc/shadow' /path/to/skill/

# C5: Crypto addresses (flag for manual review — high false positive rate)
grep -rn -E '0x[a-fA-F0-9]{40}' /path/to/skill/

Credential Access

High
Category
Privilege Escalation
Confidence
95% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 66)May include surrounding context.

md
grep -rn -E 'curl\s.*\|\s*(ba|da)sh|wget\s.*\|\s*(ba|da)sh|eval\s.*\$\(curl|pip\s+install\s+https?://' /path/to/skill/

# C4: Sensitive paths (path-format only, avoid false positives)
grep -rn -E '/\.ssh/|/\.aws/credentials|/etc/passwd|/etc/shadow' /path/to/skill/

# C5: Crypto addresses (flag for manual review — high false positive rate)
grep -rn -E '0x[a-fA-F0-9]{40}' /path/to/skill/

Credential Access

High
Category
Privilege Escalation
Confidence
90% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · references/security.md (reported line 66)May include surrounding context.

md
| C1 | Credential harvesting: reads API keys/tokens in code AND makes network calls | Both must be present in same file |
| C2 | Obfuscated payloads: base64-decoded URLs or commands | Verify decoded content is actually malicious |
| C3 | Remote execution: curl\|bash, wget\|sh, eval $(curl), pip install from HTTP URLs | |
| C4 | Sensitive filesystem access in executable code | ~/.ssh, ~/.aws/credentials, /etc/passwd, /etc/shadow |
| C5 | Crypto wallet addresses with network calls | Verify: manual review required, high FP rate |
| C6 | Symlinks targeting system paths (/etc, /usr, ~/.ssh, ~/.aws) | |
| C7 | Prompt injection: "ignore instructions", role hijacking, hidden HTML/zero-width directives | Check SKILL.md thoroughly |

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/security.md (reported line 70)May include surrounding context.

md
| C5 | Crypto wallet addresses with network calls | Verify: manual review required, high FP rate |
| C6 | Symlinks targeting system paths (/etc, /usr, ~/.ssh, ~/.aws) | |
| C7 | Prompt injection: "ignore instructions", role hijacking, hidden HTML/zero-width directives | Check SKILL.md thoroughly |
| C8 | Privilege escalation: sudo, chmod 777, setuid, writes to system paths | In executable code, not docs |
| C9 | Data exfiltration: known exfil endpoints + outbound network calls | |
| C10 | Time bombs: date/time logic paired with destructive operations | Manual review only — no grep shortcut |
| C11 | Non-asset executables in assets/ (.sh, .py, .exe, .bat, .ps1) | |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list is very broad, including generic phrases like 'evaluate skill' and 'rate skill quality', which can cause the skill to activate for ordinary review or scoring requests not specifically intended for this tool. Over-broad activation increases the chance of unintended instruction injection into unrelated workflows and can override a user's expected response style or task routing.

Content

No source excerpt is available for this finding.

Unrestricted Tool Access

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill grants unrestricted tool access without appropriate constraints. An agent with unfettered tool access can perform arbitrary actions including file modification, network requests, and code execution.

Content

Scanner excerpt · references/security.md (reported line 16)May include surrounding context.

md
3. Check assets/ for non-asset files: `find /path/to/skill/assets -type f \( -name "*.sh" -o -name "*.py" -o -name "*.exe" -o -name "*.bat" -o -name "*.ps1" \)`
4. Read SKILL.md (highest risk for prompt injection — watch for zero-width characters, HTML comments, hidden directives)
5. Grep scripts/ and all other files for patterns below
6. **Do NOT execute any scripts**

## Grep Commands

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · references/security.md (reported line 70)May include surrounding context.

md
| C5 | Crypto wallet addresses with network calls | Verify: manual review required, high FP rate |
| C6 | Symlinks targeting system paths (/etc, /usr, ~/.ssh, ~/.aws) | |
| C7 | Prompt injection: "ignore instructions", role hijacking, hidden HTML/zero-width directives | Check SKILL.md thoroughly |
| C8 | Privilege escalation: sudo, chmod 777, setuid, writes to system paths | In executable code, not docs |
| C9 | Data exfiltration: known exfil endpoints + outbound network calls | |
| C10 | Time bombs: date/time logic paired with destructive operations | Manual review only — no grep shortcut |
| C11 | Non-asset executables in assets/ (.sh, .py, .exe, .bat, .ps1) | |