The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the dxs (Datex CLI) and jq tools to manage configuration lifecycles.
Evidence: Commands such as dxs configuration get, dxs configuration validate, dxs configuration upsert, and jq .json envelope.json > body.json are used to retrieve, process, and update platform data.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and modifying external JSON configuration data that could contain malicious instructions.
Ingestion points: Data is retrieved into envelope.json via the dxs configuration get command.
Boundary markers: No specific delimiters are used to isolate ingested data from agent instructions, though the data format is structured JSON.
Capability inventory: The skill possesses the ability to write and update platform configurations via dxs configuration upsert.
Sanitization: The instructions recommend running dxs configuration validate to check the schema and integrity of the modified JSON before pushing changes.
[COMMAND_EXECUTION]: The skill involves the dynamic generation of TypeScript expressions and logic flows (ExecuteCodeActivity) to be embedded within the form configurations.
Evidence: Detailed instructions and references (e.g., references/forms.md) describe how to author code strings for field validation, initialization, and toolbar actions which are executed by the platform runtime.

form-creator