The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes user-provided component files (JSON) which represents an indirect prompt injection surface. Malicious data within these files could attempt to bypass the validation logic or influence the sub-agent's findings.
Ingestion points: Target component JSON files (e.g., body.json) provided during the audit workflow.
Boundary markers: The sub-agent prompt does not specify the use of clear delimiters or instructions to ignore embedded commands within the processed JSON content.
Capability inventory: The sub-agent is granted Read, Grep, and Glob access to the local environment; the orchestrator utilizes dxs (Datex Studio CLI) and jq tools.
Sanitization: There is no evidence of string sanitization or escaping for the data extracted from the component files before it is processed by the sub-agent.
[COMMAND_EXECUTION]: The skill uses local command-line tools to fetch and process component data as part of its standard operational workflow.
Evidence: The instructions include execution patterns for dxs source explore, dxs configuration get, and jq to extract and prepare the component body for auditing.

component-validator