youtube-transcript
Verdict safe, but 1 finding (1 medium), mainly data exfiltration. Most recent scan 2026-02-07.
The youtube-transcript skill is MODERATE RISK with primarily documentation and transparency issues rather than active security threats. The skill's core functionality (fetching YouTube transcripts via residential IP proxy) is legitimate, but lacks proper metadata declarations. Key concerns: (1) Missing allowed-tools declaration for Python/Bash usage, (2) Network communication not explicitly disclosed in description, (3) Short subprocess timeouts may cause availability issues, (4) Missing license/compatibility metadata. No evidence of data exfiltration, prompt injection, credential theft, or malicious code execution. The VPN configuration requirement is clearly documented and serves a legitimate technical purpose (bypassing YouTube's cloud IP blocks). The skill operates transparently within its stated scope.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Fetch and summarize YouTube video transcripts. Use when asked to summarize, transcribe, or extract content from YouTube videos. Handles transcript fetching via residential IP proxy to bypass YouTube's cloud IP blocks.
- Package
- 3 files, 8.4 KB
- SKILL.md
- 125 words, 5 sections
- Code
- Python (1)
SKILL.md covers: YouTube Transcript · Quick Start · Workflow · Language Codes · Setup & Configuration
Findings
MEDIUM Missing allowed-tools Declaration Data Exfiltrationllm
The skill does not declare allowed-tools in its YAML manifest. While this field is optional per the agent skills spec, its absence means users cannot verify which agent capabilities (Read, Write, Bash, Python, etc.) the skill intends to use. The skill executes Python scripts and Bash commands (wg-quick, subprocess calls) without declaring these permissions upfront.
SKILL.md
YAML manifest missing 'allowed-tools' field. Script uses: subprocess.run(['wg-quick', 'up', VPN_INTERFACE]), subprocess.run(['wg', 'show', VPN_INTERFACE]), and Python execution.What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 1.3 KB |
references/SETUP.md | doc | 2.5 KB |
scripts/fetch_transcript.py | script | 4.5 KB |
External hosts referenced
www.youtube.com noembed.com
Hostnames that appear in the skill's files. Referencing a host is not a finding; compare against the findings above and the skill's stated purpose.
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/youtube-transcript
- Raw scan result: youtube-transcript-scan.json
- Badge for your README:
- Request a re-scan