The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the installation of the @membranehq/cli package via NPM. This is the official command-line tool provided by the vendor (Membrane) to facilitate secure integration.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the Membrane CLI for authentication (membrane login), connection management (membrane connect), and executing actions (membrane action run). This is the primary method of operation and does not involve arbitrary or unsafe command construction.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Bexio records (such as contact notes or invoice descriptions) which are then processed by the agent. While this presents a potential surface for indirect prompt injection, it is a standard risk for CRM integrations and the skill provides no high-privilege capabilities that would lead to escalation.
Ingestion points: Data returned from membrane action run (e.g., contact details, invoices).
Boundary markers: Not explicitly defined in the prompt templates.
Capability inventory: Limited to membrane CLI subprocess calls for data retrieval and modification.
Sanitization: Relies on the agent's native content filtering and the Membrane platform's schema validation.

bexio