The Agent Skills Directory

[SAFE]: The skill implements a robust security-focused workflow for migrating legacy Qlik Sense extensions. It enforces strict host-isolation rules to ensure extensions do not depend on the host page's APIs or frameworks, which minimizes the risk of breaking or being compromised when the host environment changes.
[SAFE]: The included migration and verification scripts (inventory.mjs, check-no-host-apis.mjs, check-package.mjs, etc.) are read-only tools that perform static analysis via regex patterns. They do not execute the analyzed code, make network calls, or write to the filesystem outside of the specified project directory.
[SAFE]: The instructions mandate that all external resources (libraries, fonts, images) must be bundled within the extension archive. This practice prevents unauthorized run-time requests to external CDNs and avoids potential security risks from mutable remote code.
[SAFE]: The skill requires explicit user approval before performing any side-effect operations, such as installing npm packages or uploading extensions to a Qlik tenant via CLI tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes legacy source code provided by the user, which is an inherent indirect prompt injection surface for a migration tool. However, the risk is mitigated by the use of specialized static analysis scripts and a structured workflow.
Ingestion points: Legacy extension source code (JS, HTML, CSS, QEXT) read from a user-provided directory.
Boundary markers: None explicitly used; the agent is instructed to read files completely for inventory.
Capability inventory: The agent can read, write, and edit files, and execute shell commands via bash (npm, npx, zip).
Sanitization: The skill employs regex-based static inventory and compatibility checks to analyze the code structure without executing the logic of the source files.

qlik-extension-nebula-migration