Rp1
Medium
- Category
- MCP Rug Pull
- Confidence
- 70% confidence
- Finding
- npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.
Security audit
BulkTranscripts YouTube
Security checks for vulnerabilities and agentic risk
Overview
The skill is a straightforward transcript/search helper that sends YouTube-related inputs to BulkTranscripts, with visible install and privacy caveats but no hidden or destructive behavior.
Before installing, review the SKILL.md you are adding, prefer a pinned or packaged source when possible, and only set BULKTRANSCRIPTS_API_KEY if you trust BulkTranscripts with transcript requests tied to your account. Expect YouTube URLs, video IDs, channel handles, playlist IDs, and search terms to be sent to bulktranscripts.co, and approve bulk transcript fetching because it can spend credits.
Vulnerability Patterns
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Rogue AgentSelf-Modification, Session Persistence
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)
Skill Enumeration
Medium
- Category
- Agent Snooping
- Content
**Claude Code** ```bash mkdir -p ~/.claude/skills/youtube-transcripts curl -fsSL https://bulktranscripts.co/skill.md \ -o ~/.claude/skills/youtube-transcripts/SKILL.md ```
- Confidence
- 85% confidence
- Finding
- Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
Session Persistence
Medium
- Category
- Rogue Agent
- Content
**Claude Code** ```bash mkdir -p ~/.claude/skills/youtube-transcripts curl -fsSL https://bulktranscripts.co/skill.md \ -o ~/.claude/skills/youtube-transcripts/SKILL.md ```
- Confidence
- 76% confidence
- Finding
- The README instructs users to persistently install a remotely fetched `SKILL.md` into the agent's long-lived skills directory using `curl -fsSL https://bulktranscripts.co/skill.md -o .../SKILL.md`. Because the fetched content is mutable and stored for future sessions, a compromised server, DNS/TLS interception, or later upstream modification could turn this into durable agent behavior modification across sessions.
Skill Enumeration
Medium
- Category
- Agent Snooping
- Content
```bash mkdir -p ~/.claude/skills/youtube-transcripts curl -fsSL https://bulktranscripts.co/skill.md \ -o ~/.claude/skills/youtube-transcripts/SKILL.md ``` **OpenAI Codex**
- Confidence
- 80% confidence
- Finding
- Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
Skill Enumeration
Medium
- Category
- Agent Snooping
- Content
```bash mkdir -p ~/.claude/skills/youtube-transcripts curl -fsSL https://bulktranscripts.co/skill.md \ -o ~/.claude/skills/youtube-transcripts/SKILL.md ``` **OpenAI Codex**
- Confidence
- 80% confidence
- Finding
- Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
Skill Enumeration
Medium
- Category
- Agent Snooping
- Content
**OpenAI Codex** ```bash mkdir -p ~/.codex/skills/youtube-transcripts curl -fsSL https://bulktranscripts.co/skill.md \ -o ~/.codex/skills/youtube-transcripts/SKILL.md ```
- Confidence
- 85% confidence
- Finding
- Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.
Vague Triggers
Medium
- Confidence
- 90% confidence
- Finding
- The activation description says to use the skill when the user 'asks to summarize/analyze/quote a video,' 'needs YouTube research,' or 'asks what a channel posted recently.' These phrases are broad and lack explicit constraints or negative examples, which could cause unintended invocation for generic research or summarization requests.
Missing User Warnings
Low
- Confidence
- 87% confidence
- Finding
- The markdown instructs the agent to keep seen video IDs in a scratch file under the user's home cache directory. This is a user-data-affecting behavior, but the description does not explicitly warn that local state will be created and updated on disk.
Missing User Warnings
Low
- Confidence
- 83% confidence
- Finding
- The skill description tells the agent to call an external API using a YouTube video identifier, which is a network operation that transmits user-provided input to a third party. The markdown does not include any warning about this external request, privacy implications, or that transcript retrieval depends on a non-local service.
Static analysis
No suspicious patterns detected.