Security audit

Digital Sanctuary Cartographer

Security checks for vulnerabilities and agentic risk

Overview

The skill reports successful security checks and protected exports without implementing them, making its assurances unreliable.

Treat this as an unfinished planning aid. Do not rely on its PASS results, encryption claims, reports, or export messages as evidence that checks succeeded or files exist. Verify controls independently before using it for deployment or incident decisions.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

MCP server referenced without pinned version: 'npx clawhub'.

Medium

Category: MCP Rug Pull
Confidence: 97% confidence
Finding: The installation command uses npx clawhub@latest, which executes a mutable package version rather than a reviewed, pinned release; a compromised or malicious future release could execute code with the installing user's privileges. Although the finding incorrectly describes the CLI as an MCP server, the underlying supply-chain exposure is real, and the skill's local-only runtime claims do not protect its installation process.
Content: No source excerpt is available for this finding.

The script claims to run boundary stress tests but only prints predetermined security results.

Medium

Category: Excessive Agency
Confidence: 99% confidence
Finding: The script prints hardcoded security conclusions without testing consent gates, permission boundaries, data handling, or encryption, and references a report it never generates. Presenting this as a boundary stress-test command can cause users or automated agents to trust nonexistent validation and overlook real vulnerabilities; however, the file itself does not demonstrate a permission bypass or data leak.
Content: No source excerpt is available for this finding.

The script claims to export an atlas but only prints export messages; no JSON capsule is created.

Low

Category: Excessive Agency
Confidence: 99% confidence
Finding: The script reports a successful export and names a JSON file without creating any artifact; it also prints a sealing message without performing cryptographic sealing. In an export workflow, these misleading success messages could cause a user or agent to rely on a nonexistent backup or integrity protection, although this file alone shows no data theft or destructive action.
Content: No source excerpt is available for this finding.