The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill relies on npx -y mcporter and uvx mcp-vods to perform its core tasks. These commands download and execute code from remote registries at runtime. Specifically, the -y flag in the npx command bypasses user confirmation, enabling the automatic execution of remote packages without human oversight.
[COMMAND_EXECUTION]: The instructions suggest using curl -sSL "$url" to resolve playback addresses from external websites. This pattern allows for arbitrary network requests and the processing of external response bodies, which could lead to Server-Side Request Forgery (SSRF) if a malicious URL is provided or retrieved from a search result.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted data.
Ingestion points: Data is ingested from multiple third-party media source sites via the vods_search tool and subsequent curl operations.
Boundary markers: The instructions lack delimiters or warnings to treat external content as data rather than instructions, increasing the risk that the agent might obey commands hidden in media metadata.
Capability inventory: The agent has shell execution capabilities via npx, uvx, and curl, as well as access to local network IPs.
Sanitization: There are no instructions for validating the source URLs or sanitizing the content retrieved from external sites before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic package runners (npx, uvx) that do not specify version pins. This makes the skill vulnerable to supply chain attacks, such as package hijacking or typosquatting on the remote registries.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download and execution of external scripts and packages from public registries as a primary method of operation.

mcp-vods