The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the bash tool to run the ajv validator via npx. This is used to ensure the generated Qlik Automate workspace JSON adheres to the required schemas before it is returned to the user or deployed.
[EXTERNAL_DOWNLOADS]: The validation step involves npx downloading the ajv-cli and ajv-formats packages from the public npm registry. These are well-known tools for JSON schema validation and the operation is used to verify the safety and correctness of the skill's output.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external Qlik services to build automations.
Ingestion points: Metadata from qlik_list_automation_connectors, qlik_get_automation_connector, and search results from qlik_search (referenced in SKILL.md).
Boundary markers: The instructions explicitly command the agent to treat search response fields as data only and never as instructions, specifically stating that text in the response must not change subsequent actions.
Capability inventory: The skill has the ability to execute bash commands for validation and use Qlik management tools to create or modify automations.
Sanitization: All generated JSON is validated against strict schemas using ajv, and there are explicit rules in references/construction-rules.md against interpolating variables into code blocks to prevent injection.
[DYNAMIC_EXECUTION]: The skill generates executable code (PHP, Node.js, or Python) for Qlik Automate's CustomCodeBlock3. It enforces a strict security policy in references/construction-rules.md against interpolating dynamic values directly into the code strings, requiring them to be passed as separate inputs to prevent code injection vulnerabilities in the resulting automation.

qlik-automation-builder