ClawHub
Back to skill

Security audit

Sentinel Vanguard AI Skill Security Auditor

Security checks for vulnerabilities and agentic risk

Overview

This skill is a read-only auditor for pasted AI-skill text and does not include code, installers, network use, credential access, or persistence.

Install only if you want an agent-facing security review helper for pasted skill content. It should not be used as a live scanner for URLs or local repositories because its own instructions intentionally avoid network access, code execution, and filesystem inspection.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (7)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
70% 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/l1-rules.md (reported line 18)May include surrounding context.

md
| L1-D07 | `DELETE FROM` without WHERE | HIGH | Bulk database record deletion |
| L1-D08 | `DROP TABLE` | CRITICAL | Schema destruction |
| L1-D09 | `TRUNCATE` | HIGH | Table data wipe |
| L1-D10 | `git push --force` | MEDIUM | History-rewriting push |

## Exfiltration Signals (L1-E series)

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · references/l1-rules.md (reported line 31)May include surrounding context.

md
| L1-E05 | `process.env.` | HIGH | Node.js env var access |
| L1-E06 | `dotenv` combined with HTTP call | HIGH | Secret loading + egress = red flag |
| L1-E07 | Undeclared `http://` or `https://` | LOW | Undocumented external call |
| L1-E08 | `keytar` / `keychain` | HIGH | OS keychain access |
| L1-E09 | `clipboard.read` | MEDIUM | Clipboard harvesting |
| L1-E10 | `screenshot` / `captureScreen` | HIGH | Screen capture |

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/l1-rules.md (reported line 60)May include surrounding context.

md
| L1-P04 | Wildcard scopes `*` | HIGH | Unrestricted access scope |
| L1-P05 | Disabling audit/log | HIGH | Telemetry suppression |
| L1-P06 | `sudo` | HIGH | Superuser command execution |
| L1-P07 | `chmod 777` | MEDIUM | World-writable permissions |
| L1-P08 | `setuid` / `setgid` | CRITICAL | Privilege escalation |

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest description says the skill should be used whenever a user wants to "audit, review, vet, or assess" safety and security, and extends this to "any request to evaluate trustworthiness of agent-facing code." This activation scope is broad and open-ended, with no exclusion conditions or negative examples, increasing the chance of matching ordinary review requests beyond the intended narrow skill context.

Content

No source excerpt is available for this finding.

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/l1-rules.md (reported line 60)May include surrounding context.

md
| L1-P04 | Wildcard scopes `*` | HIGH | Unrestricted access scope |
| L1-P05 | Disabling audit/log | HIGH | Telemetry suppression |
| L1-P06 | `sudo` | HIGH | Superuser command execution |
| L1-P07 | `chmod 777` | MEDIUM | World-writable permissions |
| L1-P08 | `setuid` / `setgid` | CRITICAL | Privilege escalation |

Unsafe Defaults

Medium
Category
Tool Misuse
Confidence
70% confidence
Finding

Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.

Content

Scanner excerpt · references/l1-rules.md (reported line 60)May include surrounding context.

md
| L1-P04 | Wildcard scopes `*` | HIGH | Unrestricted access scope |
| L1-P05 | Disabling audit/log | HIGH | Telemetry suppression |
| L1-P06 | `sudo` | HIGH | Superuser command execution |
| L1-P07 | `chmod 777` | MEDIUM | World-writable permissions |
| L1-P08 | `setuid` / `setgid` | CRITICAL | Privilege escalation |

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill presents a Chinese-language motto in the instructions without offering a language preference or documenting a locale-specific purpose. This can conflict with language/locale policy expectations when a skill introduces a specific language by default without user opt-in.

Content

No source excerpt is available for this finding.