GLM-V-Grounding

Security checks across malware telemetry and agentic risk

Overview

This appears to be a purpose-aligned GLM-V image/video grounding skill, but it uses a Zhipu API key and can send user-provided media to an external provider.

Install only if you are comfortable using a Zhipu API key and sending selected images/videos to the Zhipu GLM-V service. Use a dedicated API key, protect the generated .env file, run dependencies in a virtual environment, and avoid pointing the skill at sensitive local files unless you intend them to be processed externally.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

ASI02: Tool Misuse and Exploitation

Medium

What this means

Images or videos you provide may be read and processed by the skill.

Why it was flagged

The skill is explicitly allowed to read user-provided local media paths or fetch remote URLs, which is expected for visual grounding but can expose private media if the user points it at sensitive files.

Skill content

URL/local file handling: the skill can read local files or fetch user-provided URLs for processing/visualization; URL inputs are restricted to public http/https targets

Recommendation

Only provide files and URLs you intend to process, and confirm local paths before using the skill on private media.

ASI03: Identity and Privilege Abuse

Medium

What this means

Anyone who can read the skill's .env file could obtain the API key and use the associated Zhipu account quota or permissions.

Why it was flagged

The setup helper stores the Zhipu API key in a local .env file. This is expected for the integration, but it is still a sensitive credential.

Skill content

env_vars["ZHIPU_API_KEY"] = api_key
...
f.write(f"{key}={value}\n")

Recommendation

Use a dedicated, revocable API key; keep the .env file private; and avoid committing it to version control.

ASI07: Insecure Inter-Agent Communication

Medium

What this means

Content you ask the skill to analyze may be sent to Zhipu's API for processing.

Why it was flagged

The CLI sends request payloads to the external Zhipu Chat Completions endpoint. This matches the skill purpose, but it means prompts and media-derived payloads leave the local environment.

Skill content

DEFAULT_API_URL = "https://open.bigmodel.cn/api/paas/v4/chat/completions"
...
resp = requests.post(api_url, json=payload, headers=headers, timeout=timeout)

Recommendation

Review the provider's data handling terms and avoid sending sensitive or regulated media unless that use is allowed.

ASI04: Agentic Supply Chain Vulnerabilities

Low

What this means

A future dependency version could change behavior or introduce a vulnerability.

Why it was flagged

Dependencies are specified with minimum versions rather than exact pins or hashes. This is common for Python skills but means installed versions may vary over time.

Skill content

requests>=2.31.0
Pillow>=10.0.0
opencv-python>=4.8.0
numpy>=1.24.0
matplotlib>=3.7.0
decord>=0.6.0

Recommendation

Install in an isolated virtual environment and consider pinning exact package versions or using a lockfile for repeatable installs.