Back to skill

Security audit

Historical Map

Security checks for vulnerabilities and agentic risk

Overview

This skill coherently generates historical map images from user-selected map data and textures, with no evidence of hidden persistence, exfiltration, or deceptive behavior.

Install this in an isolated Python environment, prefer pinned dependency versions if reproducibility matters, and only process GeoJSON, event JSON, and image texture files from sources you trust. Avoid extremely large basemap or parchment images unless the runtime has memory and CPU limits.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:8
Finding

Unpinned Third-Party Dependencies Create a Supply-Chain Risk

Content
View full analysis
=0.14", "matplotlib>=3.8", "numpy>=1.24", "Pillow>=10.0", "pyproj>=3.6", "shapely>=2.0" ] }, "install": [ { "id": "pip-install", "kind": "pip", "packages": [ "geopandas>=0.14", "matplotlib>=3.8", "numpy>=1.24", "Pillow>=10.0", "pyproj>=3.6", "shapely>=2.0" ], "label": "Install map generation dependencies" } ] ``` The documentation also recommends installing dependencies without any version constraints: ```bash pip install geopandas matplotlib numpy Pillow pyproj shapely ``` ### Technical Analysis The Skill permits installation of any package version equal to or newer than the specified minimum version. The documented installation command removes even those minimum constraints. No lock file, integrity hashes, or trusted package-index restrictions are provided. Python package installation can execute package build and installation logic. Consequently, the effective code installed by the Skill can change after the Skill itself has been reviewed. A compromised package release, malicious transitive dependency, dependency-confusion condition, or compromised package source could therefore introduce attacker-controlled code. There is no evidence that any currently named dependency is malicious. The vulnerability is the absence of reproducible, integrity-verified dependency resolution. ### Attack Path 1. An attacker compromises a permitted future release of a direct or transitive dependency, publishes a malicious package through a package source available to pip, or influences the configured package index. 2. A user or Agent follows the Skill installation metadata or documented `pi ...[truncated 1011 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
generate.py:280
Finding

Pillow Decompression-Bomb Protection Is Effectively Disabled for User-Selected Images

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
79% confidence
Finding

The skill documentation describes code usage that reads local files and user-supplied paths such as GeoJSON, textures, and event JSON, but it does not declare any explicit tool scope or file access boundaries. In an agent environment, undeclared file-read capability can lead to overbroad access, unclear consent, and accidental reading of sensitive local files if the runtime infers capabilities from the examples or implementation.

Content

No source excerpt is available for this finding.

Dynamic import via __import__()

Medium
Category
Dangerous Code Execution
Confidence
75% confidence
Finding

Dynamic import() can load arbitrary modules at runtime, bypassing static analysis and potentially importing malicious code.

Content

Scanner excerpt · generate.py (reported line 25)May include surrounding context.

python
missing = []
    for mod in ['geopandas', 'matplotlib', 'numpy', 'PIL', 'pyproj', 'shapely']:
        try:
            __import__(mod)
        except ImportError:
            missing.append(mod)
    if missing:

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The preset event labels and captions are embedded in Chinese, and the generated output also includes Chinese text fields as part of the default content. Because the file does not present this as an explicit opt-in locale choice or region-specific requirement, it imposes a language preference in a way that can violate organizational language/locale policy.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · generate.py (reported line 152)May include surrounding context.

python
break

    def _get_projection_transformer(self, projection, region):
        """Create a projection transformer."""
        if projection == 'platecarree':
            return None  # No projection needed

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The usage example introduces a --title-cn parameter specifically for Chinese text, and the parameters section later describes it as a Chinese subtitle. This suggests a built-in language-specific assumption without clarifying whether other languages are supported or whether Chinese labeling is optional by user choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The code explicitly searches for Chinese fonts and conditionally renders Chinese subtitles and timeline text, reinforcing a built-in locale preference. Without an explicit language selection flow or justification that the skill is intended for Chinese-language output, this is a natural-language locale policy concern.

Content

No source excerpt is available for this finding.