Back to skill

Security audit

YouTube Watcher

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple YouTube transcript helper with some quality and supply-chain caveats, but no artifact-backed malicious or deceptive behavior.

Install only if you are comfortable relying on yt-dlp from your configured Homebrew or pip sources, and expect the skill may need repair because the referenced transcript script is not included in the inspected package. Treat non-YouTube video requests carefully because the triggers are broader than the stated purpose.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • 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 (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:12
Finding
Unpinned Third-Party Dependency Installation## Vulnerability Details **File Location**: `SKILL.md`, line 12 **Vulnerability Type**: `T08: Insecure Dependencies` **Risk Level**: Medium **Vulnerable Code Snippet**: ```yaml metadata: {"clawdbot":{"emoji":"📺","requires":{"bins":["yt-dlp"]},"install":[{"id":"brew","kind":"brew","formula":"yt-dlp","bins":["yt-dlp"],"label":"Install yt-dlp (brew)"},{"id":"pip","kind":"pip","package":"yt-dlp","bins":["yt-dlp"],"label":"Install yt-dlp (pip)"}]}} ``` ### Technical Analysis The installation metadata permits `yt-dlp` to be installed from a third-party package registry without specifying an audited version or an integrity hash. Consequently, the installed artifact can change after this Skill has been reviewed. This does not demonstrate that the legitimate `yt-dlp` package is malicious. The risk arises because installation security depends on the future state of the package registry, publisher account, configured package source, and dependency resolution process. If any of those supply-chain components are compromised or redirected, an attacker-controlled package release could be installed. The Homebrew installation alternative is also unversioned, although its distribution and verification controls differ from those of Python package installation. ### Attack Path 1. An attacker compromises a relevant package publisher or distribution account, package source, or dependency delivery channel. 2. The attacker publishes or serves a malicious release under the expected `yt-dlp` package identity. 3. A user or agent installs the Skill dependency through the declared unpinned installation method. 4. The resolver selects the attacker-controlled release because no approved version or artifact hash is enforced. 5. Malicious package installation behavior or subsequently invoked package code executes with the privileges of the installing or invoking user. This exploitation path is conditional on compromise or manipulation of the upstrea ...[truncated 725 chars]
Remediation
## Remediation Suggestions 1. Pin `yt-dlp` to a specifically reviewed version rather than resolving the latest available release. 2. For Python installations, distribute a requirements or lock file containing cryptographic hashes and install with hash verification, such as `pip install --require-hashes`. 3. Configure an explicitly trusted package index and disable unintended fallback indexes to reduce dependency-confusion and source-redirection risks. 4. For Homebrew installations, document the trusted tap and use an approved versioned formula or verified package artifact where feasible. 5. Regularly review pinned dependency versions for security advisories and update them through a controlled review process. 6. Install and execute the dependency as an unprivileged user in a restricted environment with only the filesystem and network access required for transcript retrieval.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger phrases include very generic terms such as "summarize video" and "analyze video", which can match many ordinary requests that are not clearly limited to YouTube transcript retrieval. This can cause the skill to activate in unintended contexts, leading the agent to invoke external tooling or fetch remote content when another safer or more appropriate capability should have handled the request.

Static analysis

No suspicious patterns detected.