The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @membranehq/cli package via npm. This is a legitimate dependency provided by the platform vendor (Membrane) to facilitate secure authentication and interaction with connectors.\n- [COMMAND_EXECUTION]: The skill uses the membrane CLI to execute actions and manage connections. These commands are part of the intended functionality for interacting with the Moskit integration and involve standard CLI operations.\n- [PROMPT_INJECTION]: The skill retrieves and processes data from external actions, which introduces a surface for indirect prompt injection.\n
Ingestion points: Untrusted data enters the agent context through the output of membrane action list (JSON schemas) and membrane action run (action output from the Moskit service).\n
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat retrieved content as untrusted data rather than instructions.\n
Capability inventory: The skill possesses the capability to execute shell commands (via the membrane CLI), perform network operations, and write data back to the Moskit service.\n
Sanitization: No sanitization, escaping, or validation of the content retrieved from external sources is described in the instructions.\n- [PROMPT_INJECTION]: There is an inconsistency in the skill's metadata. The body of the SKILL.md file describes Moskit as a "session replay and product analytics tool," whereas the YAML frontmatter and the provided overview indicate it is a CRM for managing organizations, activities, and tasks. This discrepancy could lead to the agent misinterpreting the sensitivity or purpose of the data it processes.

moskit