Security audit

smart-files

Security checks for vulnerabilities and agentic risk

Overview

This local file-management skill is mostly disclosed, but it needs Review because its workspace boundary checks can be bypassed to read process-readable files outside the intended workspace.

Install only if you are comfortable with a local tool that recursively reads files and can show raw content snippets. Keep use to trusted workspaces, avoid directories containing secrets, do not use --force unless you intentionally want external paths scanned, and treat watch journaling as persistent metadata about file activity.

A.I.G

Vulnerability Patterns

Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
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

Findings (1)

T05 · Unauthorized Access and Privilege Escalation

Warning

Location: smart-files.js:18
Finding: Workspace Boundary Bypass Enables Unauthorized External File Access
Content: View full analysis
0) { const ext = path.extname(entry.name).toLowerCase(); if (!extensions.includes(ext)) continue; } files.push({ path: fullPath, name: entry.name, size: fs.statSync(fullPath).size }); } } return files; } ``` The `--info` implementation accepts an arbitrary path without calling `resolveDir` at all: ```js function fileInfo(filepath) { try { const stat = fs.statSync ...[truncated 5627 chars]
Remediation: View remediation

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
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (4)

Tool Parameter Abuse

High

Category: Tool Misuse
Confidence: 85% 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 · README.md (reported line 247)May include surrounding context.
cp memory/smart-files-journal.json memory/smart-files-journal.backup && > memory/smart-files-journal.json

Or remove entirely

rm -f memory/smart-files-journal.json

text
### What's NOT Protected

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 · smart-files.js (reported line 242)May include surrounding context.
js
const files = collectFiles(searchRoot); const categories = { code: { exts: ['.js', '.ts', '.py', '.html', '.css', '.json', '.yaml', '.yml', '.xml', '.sh', '.bash', '.zsh', '.rb', '.go', '.rs', '.c', '.cpp', '.h', '.java', '.kt', '.swift', '.php', '.sql', '.md', '.txt'], dir: 'code' }, data: { exts: ['.csv', '.tsv', '.json', '.jsonl', '.xml', '.yaml', '.yml', '.toml', '.ini', '.cfg', '.conf', '.env', '.sql'], dir: 'data' }, docs: { exts: ['.pdf', '.doc', '.docx', '.rtf', '.odt', '.tex', '.epub'], dir: 'docs' }, media: { exts: ['.jpg', '.jpeg', '.png', '.gif', '.svg', '.webp', '.bmp', '.mp4', '.avi', '.mov', '.mkv', '.mp3', '.wav', '.flac', '.ogg', '.m4a'], dir: 'media' }, archives: { exts: ['.zip', '.tar', '.gz', '.rar', '.7z', '.bz2', '.xz'], dir: 'archives' },

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 84% confidence
Finding: The description advertises broad, loosely constrained behavior such as content-aware file reading, snippet printing, continuous monitoring, and use of flags that can suppress visibility or override workspace boundaries. In an agent skill, this kind of activation language can encourage invocation in more contexts than necessary and increases the chance of overbroad file access or unintended disclosure of sensitive content.
Content: No source excerpt is available for this finding.

Intent-Code Divergence