The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the @membranehq/cli package via npm. This is the official utility for the Membrane platform, provided by the skill's author.
[COMMAND_EXECUTION]: The skill facilitates the execution of various commands using the membrane CLI to manage authentication, connections, and CRM records. This includes membrane login, membrane connect, and membrane action run.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection (Category 8) due to how it handles natural language inputs and dynamic actions.
Ingestion points: Untrusted data enters the context through user-provided queries in membrane action list --intent "QUERY", action descriptions in membrane action create "DESCRIPTION", and JSON parameters in membrane action run --input.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to distinguish between its own logic and potentially malicious instructions embedded in the intent strings or CRM data.
Capability inventory: The skill allows the agent to create and execute actions that can read from or write to the connected Pancake CRM account.
Sanitization: There is no evidence of input validation or sanitization for the strings passed to the CLI flags.

pancake-crm