The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits significant metadata poisoning through contradictory information. The name 'Conveyor' is used, but the description describes a packaging tool (conveyor.dev), the action list describes a compliance platform (conveyor.com), and the documentation link points to an unrelated transit service (conveyal.com). This misleading information can result in the agent or user interacting with unintended services or APIs.\n- [COMMAND_EXECUTION]: The skill relies on the agent executing shell commands through the Membrane CLI (membrane). These commands are used for lifecycle management, including logging in, creating connections, and executing API actions. This is the primary mechanism for the skill's operation.\n- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli Node.js package. This is a vendor-controlled tool required to use the integration. While it is from a known entity related to the author, it is an external dependency that must be installed globally.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its ingestion of external API data. Malicious content within Conveyor API responses could potentially influence the agent's logic.\n
Ingestion points: Untrusted data is retrieved via membrane action run and membrane request commands.\n
Boundary markers: There are no delimiters or instructions provided to the agent to treat API output as untrusted data.\n
Capability inventory: The agent possesses the capability to run CLI commands and perform network requests via the proxy.\n
Sanitization: No sanitization or validation of external API content is mentioned or implemented in the instructions.

conveyor