The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the membrane command-line utility to facilitate authentication and action execution. These commands, such as membrane action run and membrane connect, are standard operations for the vendor's integration framework and do not include unauthorized command execution.
[EXTERNAL_DOWNLOADS]: The skill guides the installation of the @membranehq/cli npm package. This is an official tool from the vendor organization, and its installation is required for the intended functionality of the integration.
[PROMPT_INJECTION]: The skill processes data from the Gong API, such as call transcripts, which constitutes a potential surface for indirect prompt injection. This is a characteristic of integrations that handle external content and is considered a baseline risk for this type of skill.
Ingestion points: Data from Gong is ingested via actions defined in SKILL.md like get-call-transcripts and get-calls-extensive.
Boundary markers: No specific delimiters or instruction-bypass warnings are defined for the processed data.
Capability inventory: The skill possesses the capability to run predefined actions, including creating or updating meetings and calls in Gong.
Sanitization: The skill relies on the standard data handling provided by the Membrane platform's action execution environment.

gong