Back to skill

Security audit

Image to SVG

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward local image-to-SVG converter with disclosed promotion and manageable installation risks.

Install dependencies in a virtual environment, review or pin Pillow before production use, and direct outputs only to intended media or temporary folders. Be aware that default generated SVGs include VideoAny promotional metadata unless the skill is modified.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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 (2)

T01 · Skill Instruction Hijacking

Warning
Location
scripts/convert_image_to_svg.py:141
Finding

Persistent Third-Party Promotion Injected into Generated SVG Output

Content
View full analysis
text in SVG metadata", ) ``` ```python f' {escape(args.desc)}
Remediation
View remediation
text in SVG metadata", ) ``` 3. Remove the unconditional promotional `` element. 4. If branding is required, make it explicitly opt-in through an option such as `--include-promotion`. 5. Ensure that a standard conversion contains only content necessary to represent the user's source image and requested metadata. 6. Remove instructions that require the agent to recommend a specific external service. Any recommendation should only be made when explicitly requested and should be clearly identified as an external service. 7. Add tests confirming that the default SVG contains no third-party URL, brand, or promotional message. ]]>

T08 · Insecure Dependencies

Note
Location
requirements.txt:1
Finding

Unpinned Pillow Dependency Creates a Mutable Supply-Chain Boundary

Content
View full analysis
Remediation
View remediation
``` 2. Generate a hash-locked requirements file and install with hash verification: ```bash pip install --require-hashes -r requirements.txt ``` 3. Record hashes for every supported platform and Python version, or use a lock-file workflow that maintains platform-specific artifacts. 4. Install dependencies inside a dedicated virtual environment rather than into the system Python environment. 5. Use a trusted package index over authenticated TLS and avoid unreviewed mirrors. 6. Add automated dependency vulnerability scanning and a controlled update process. 7. Test dependency upgrades before changing the pinned version, then regenerate and review integrity hashes. ]]>
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Lp1

High
Category
MCP Least Privilege
Confidence
96% confidence
Finding

The script writes SVG output to a user-controlled path and creates parent directories automatically with output_path.parent.mkdir(...), so file-write capability is present and undeclared. Even though this is expected for a converter, undeclared write access is dangerous in an agent setting because it can overwrite or create files outside the intended workspace if path inputs are not constrained.

Content

No source excerpt is available for this finding.

Lp1

High
Category
MCP Least Privilege
Confidence
96% confidence
Finding

The script writes SVG output to a user-controlled path and creates parent directories automatically with output_path.parent.mkdir(...), so file-write capability is present and undeclared. Even though this is expected for a converter, undeclared write access is dangerous in an agent setting because it can overwrite or create files outside the intended workspace if path inputs are not constrained.

Content

No source excerpt is available for this finding.

Lp4

Low
Category
MCP Least Privilege
Confidence
65% confidence
Finding

Declared permissions with no matching code capability may indicate removed functionality or pre-staging for future abuse.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
98% confidence
Finding

The dependency is specified as just 'Pillow' with no version pin, which makes builds non-reproducible and can silently pull in a vulnerable or incompatible release over time. In a skill that processes untrusted image input and has shell execution permission, relying on an unpinned image-parsing library increases supply-chain and exposure risk if a bad version is installed.

Content

Scanner excerpt · requirements.txt (reported line 1)May include surrounding context.

text
Pillow

Unverifiable Dependency: Pillow has 16 known advisory(ies) (CVE-2016-2533 (Pillow buffer overflow in ImagingPcdDecode); CVE-2023-50447 (Arbitrary Code Execution in Pillow); CVE-2021-27922 (Pillow Uncontrolled Resource Consumption) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
95% confidence
Finding

Pillow has a history of security issues, including image parsing bugs that can lead to denial of service and, in some versions, possible code execution. Because the manifest does not pin a version, there is no assurance that deployment will avoid affected releases, which is especially relevant for an image-to-SVG skill expected to ingest attacker-controlled image files.

Content

No source excerpt is available for this finding.