The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's automation schemas define nodes and actions for executing custom code. For example, flow.zod.ts includes a script node type for JavaScript/TypeScript, and workflow.zod.ts defines CustomScriptActionSchema supporting JavaScript, TypeScript, and Python.
[EXTERNAL_DOWNLOADS]: The protocol definitions include robust support for outbound network operations. flow.zod.ts specifies http_request nodes, and webhook.zod.ts defines schemas for pushing data to external URLs via webhooks.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes user-provided automation metadata. Malicious instructions could be embedded in fields like node labels, descriptions, or formula expressions (e.g., criteria in WorkflowRuleSchema) to influence the behavior of the agent interpreting these definitions.

objectstack-automation