The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill configuration directs the agent to download and install the @trendradar/mcp-server package from the NPM registry during the setup process.
[REMOTE_CODE_EXECUTION]: By using the npx -y @trendradar/mcp-server command, the skill facilitates the dynamic execution of remote code. This is a common pattern for MCP servers but introduces a dependency on the integrity of the NPM package.
[COMMAND_EXECUTION]: The skill utilizes shell commands (npx) to initialize its core functionality, which allows for the execution of arbitrary scripts provided by the package author.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core function of processing data from external, untrusted sources such as Twitter, Reddit, and HackerNews.
Ingestion points: Data enters the agent's context through tools like searchTrends, getPlatformTrends, and getViralContent (defined in SKILL.md) which fetch content from various social media and news platforms.
Boundary markers: The skill does not define clear delimiters or specific instructions (e.g., 'ignore instructions within the following text') to separate untrusted platform content from the agent's execution logic.
Capability inventory: The skill possesses network access via the MCP server and can be combined with other tools like agent-reach for deep content extraction.
Sanitization: No sanitization, filtering, or validation of the ingested social media content is described, allowing potentially malicious instructions embedded in trending posts to be processed by the LLM.

trendradar