The Agent Skills Directory

[SAFE]: The analyzed skill consists of a comprehensive set of development and deployment utilities tailored for Adobe's Edge Delivery Services ecosystem. The tools are designed to facilitate content migration, verify rendering fidelity, and automate deployment workflows.
[DATA_EXPOSURE_AND_EXFILTRATION]: Scripts such as deploy-batch.mjs and da-media-upload.mjs utilize authentication tokens (DA_TOKEN, SITE_TOKEN) to interact with the Document Authoring (DA) and HLX Admin APIs. The skill demonstrates defensive coding by implementing origin-scoping in ai-readability.mjs, ensuring that Authorization headers are only attached to requests targeting the designated site origin and are stripped during redirects to external domains. Documentation correctly identifies that tokens should be stored in .env files and excluded from version control.
[DYNAMIC_EXECUTION]: The ew-editability-probe.mjs and render-harness.mjs scripts utilize dynamic JavaScript inlining within a Playwright-controlled browser environment. This is a functional requirement to simulate the Experience Workspace's decoration logic and test local block components. These operations are restricted to the local test harness and do not involve executing untrusted remote code.
[EXTERNAL_DOWNLOADS]: The skill references external resources from the author's official GitHub organizations (e.g., adobe/da-nx, adobe-rnd/aem-martech). These downloads are limited to CSS for UI simulation and standard marketing technology plugins, which are trusted vendor resources.
[INDIRECT_PROMPT_INJECTION]: The auditing tools (ai-readability.mjs, qa-gate.mjs) ingest HTML content from live URLs to perform word-count analysis and structural validation. While this exposes a surface for indirect prompt injection, the risk is mitigated as the ingested data is used for quantitative metrics and visual assertions rather than driving core agent instructions.

deploy