JoeVault

• Outcome: Review
• Latest audit: Sep 16, 2026 · 4:29 AM UTC
• Version: 1.0.0

Overview

This is a coherent local cleanup skill, but its quarantine script can move data outside the intended archive path if given an unsafe label.

Review and harden scripts/quarantine_path.py before installing or using this broadly. Only run it with explicit source paths you have audited, use simple label names with no slashes or '..', avoid moving large or recently active workspaces, and keep backups because quarantine moves are persistent.

A.I.G

Vulnerability Patterns

• Insecure Skill Coding Practices: Finds exploitable flaws such as hardcoded secrets or command injection
• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally

Findings (1)

T09 · Insecure Skill Coding Practices

Warning

• Location: scripts/quarantine_path.py:27

Finding

Archive Path Traversal Through Unvalidated Label

Content

View full analysis

Vulnerability Details

File Location: scripts/quarantine_path.py, lines 27-43
Vulnerability Type: Path traversal and archive-boundary escape
Risk Level: Medium

Vulnerable Code

python

ap.add_argument('--label', required=True, help='Target folder/file name inside the archive root')
ap.add_argument('--note')
args = ap.parse_args()

source = Path(args.source)
archive_root = Path(args.archive_root)
target = archive_root / args.label
manifest = archive_root / 'MANIFEST.md'

if not source.exists() and not source.is_symlink():
    raise SystemExit(f'Source does not exist: {source}')
if target.exists():
    raise SystemExit(f'Target already exists: {target}')

archive_root.mkdir(parents=True, exist_ok=True)
shutil.move(str(source), str(target))
append_manifest(manifest, source, target, args.note)
print(f'Moved {source} -> {target}')
print(f'Updated manifest: {manifest}')

Technical Analysis

The script states that --label is a target name inside the archive root, but it accepts the argument without validating its path semantics. The expression archive_root / args.label does not enforce containment:

• An absolute label causes pathlib to discard archive_root.
• A relative label containing .. can resolve outside archive_root.
• Labels containing path separators can create unintended nested destinations.

The script only checks whether the resulting target already exists. It does not normalize the destination and verify that it remains beneath the resolved archive root. Consequently, a caller who controls command-line arguments can make the quarantine operation move a valid source to an arbitrary nonexistent destination permitted by the process's filesystem privileges.

Attack Path

1. The attacker or unsafe automation selects an existing source path that the process can move.

2. It invokes the script with an escaping label, for example:

   bash

   python3 scripts/quarantine_path.py \
     --source /path/to/source \
   

...[truncated 1244 chars]

Remediation

View remediation

Remediation Suggestions

Treat --label strictly as one filesystem entry name rather than a path:

1. Reject empty labels, absolute paths, . and ...
2. Reject labels containing /, \, or multiple path components.
3. Require Path(args.label).name == args.label.
4. Resolve the archive root and proposed target, then explicitly enforce containment before moving.
5. Perform the containment check before creating directories or modifying the source.
6. Add tests for absolute labels, traversal sequences, mixed separators, and valid ordinary labels.
7. Consider adding --dry-run and explicit confirmation options for destructive moves.

Example hardening:

python

label = Path(args.label)
if (
    not args.label
    or label.is_absolute()
    or args.label in {'.', '..'}
    or label.name != args.label
    or '/' in args.label
    or '\\' in args.label
):
    raise SystemExit('Label must be a single file or directory name')

archive_root = Path(args.archive_root).resolve()
target = (archive_root / args.label).resolve(strict=False)

if not target.is_relative_to(archive_root):
    raise SystemExit('Target must remain inside the archive root')
if target == archive_root:
    raise SystemExit('Target cannot be the archive root')
if target.exists() or target.is_symlink():
    raise SystemExit(f'Target already exists: {target}')

The source may also be normalized and checked against policy-defined protected paths before shutil.move() is called.

SkillSpector

Vulnerability Patterns

• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (1)

Undeclared Tool Scope

Medium

• Category: MCP Least Privilege
• Confidence: 85% confidence

Finding

The skill instructs the operator to run local scripts that read from and move filesystem paths, but it does not declare any explicit tool scope such as allowed-tools or permissions. That mismatch weakens reviewability and can enable broader-than-expected file access or modification if the agent runtime infers capabilities implicitly, especially because the skill deals with home directories, workspaces, and archival moves.

Content

No source excerpt is available for this finding.