The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/run.js utility is a legitimate tool used to execute MCP operations and manage local data caching. It provides the agent with controlled access to the KeyAPI LinkedIn service.\n- [DATA_EXFILTRATION]: All network operations are directed toward the vendor's official domain (mcp.keyapi.ai). The data transmission is limited to the retrieval of LinkedIn profile and activity information as per the skill's stated purpose.\n- [CREDENTIALS_UNSAFE]: The skill correctly uses an environment variable (KEYAPI_TOKEN) for authentication. The included setup script provides a secure method for users to input their token, which is then stored in a local .env file, adhering to standard development practices.\n- [PROMPT_INJECTION]: The skill ingests third-party content from LinkedIn (bios, posts, comments). While this constitutes a surface for indirect prompt injection, it is an essential component of the skill's functionality. The analysis did not reveal any exploitable paths that would allow this content to bypass the agent's primary safety instructions. Ingestion points: get_user_posts, get_user_comments, get_user_profile in scripts/run.js. Boundary markers: Not explicitly implemented in output reporting. Capability inventory: MCP tool execution and local file writing. Sanitization: JSON schema validation is performed by the MCP client.

keyapi-linkedin-user-analytics