The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @membranehq/cli package globally via NPM and also uses npx to run it. These are vendor-provided tools for the Membrane platform.
[COMMAND_EXECUTION]: All primary functions are performed by executing shell commands using the membrane CLI tool, including logging in, connecting to services, and running actions.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from external sources (Finch/Membrane records).
Ingestion points: Results from membrane action run and membrane action list are directly processed by the agent.
Boundary markers: No delimiters or safety instructions are provided to distinguish external data from agent instructions.
Capability inventory: The agent has the capability to execute shell commands via the membrane tool.
Sanitization: No sanitization or validation of the ingested external content is performed.

finch