video-stt

Security checks across static analysis, malware telemetry, and agentic risk

Overview

The skill mostly matches its video transcription purpose, but its shell wrapper can turn crafted options into local Python code execution and it installs dependencies at runtime.

Review before installing or running. Prefer the Python helper over the shell wrapper, avoid passing untrusted -o/-m/-f values, install dependencies manually from trusted sources, and do not set cloud API keys for this version because cloud mode is not implemented.

Static analysis

No static analysis findings were reported for this release.

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.

What this means

A crafted option value could run arbitrary Python commands under the user's account instead of only transcribing a video.

Why it was flagged

MODEL, FORMAT, and OUTPUT_FILE are populated from command-line arguments and interpolated into Python source code inside python3 -c without escaping or validation.

Skill content
python3 -c "... model = whisper.load_model('$MODEL') ... if '$FORMAT' == 'json': ... with open('$OUTPUT_FILE', 'w') ..."
Recommendation

Avoid using the shell wrapper with untrusted option values until it is fixed. The publisher should call a Python script with argv parameters, validate allowed model/format values, and avoid building executable code from strings.

What this means

Running the skill may change the local environment and download third-party code, which is expected for transcription tooling but should be explicit and reviewed.

Why it was flagged

The Python helper can install Homebrew packages and an unpinned Python package at runtime, while the registry metadata lists no required binaries or install spec.

Skill content
subprocess.run(["brew", "install", cmd], check=False) ... subprocess.run(["uv", "pip", "install", "whisper"], check=True)
Recommendation

Review and install dependencies yourself, prefer pinned package names/versions, and consider disabling automatic package installation in the helper.

What this means

Users may expect cloud transcription or set API keys unnecessarily, even though the provided shell implementation only supports local transcription.

Why it was flagged

The interface advertises cloud API mode but the implementation exits as not implemented.

Skill content
echo "  --api                Use cloud API" ... echo "Cloud API mode not implemented yet"
Recommendation

Do not provide cloud API credentials unless the implementation is updated and reviewed; treat cloud support as unavailable in this version.