The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses dynamic module loading based on computed file paths. The scripts capture-content.mjs, dom-equality.mjs, donor-probe.mjs, and slot-coverage.mjs all dynamically resolve and import the live-session.mjs utility using pathToFileURL. Furthermore, source-normalize.mjs dynamically imports user-defined ledger modules via the --normalize command-line argument.- [DYNAMIC_EXECUTION]: The skill requires the agent to programmatically generate and then execute custom renderer scripts (stardust/reskin/renderers/<archetype>.mjs) during Phase 4. This pattern of runtime code generation and subsequent execution increases the risk of unintended behaviors.- [DYNAMIC_EXECUTION]: The capture-content.mjs script utilizes eval() to execute the IMG_VISIBLE predicate string within the browser context during the page evaluation phase.- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from external, untrusted websites (<content-url> and --donor <url>) to build a content model. This data is then used by the agent to author a mapping brief and generate rendering logic, which provides an attack surface for indirect prompt injection.
Ingestion points: Data enters the system via page.goto in capture-content.mjs and dom-equality.mjs.
Boundary markers: None explicitly defined in the prompts; data is stored in content-model.json structures.
Capability inventory: The skill has file system access (fs.writeFileSync), browser automation capabilities, and the ability to execute generated scripts.
Sanitization: The skill employs whitespace normalization and recommends HTML-escaping in the generated renderer, but the intermediate reasoning phase lacks robust sanitization against adversarial content.- [EXTERNAL_DOWNLOADS]: The skill setup requires the installation of the impeccable dependency from a third-party GitHub repository (github.com/pbakaus/impeccable) and the playwright package via npm.- [COMMAND_EXECUTION]: The skill procedure involves several shell commands, including npm install, python3 -m http.server for local prototype serving, and the execution of multiple project-local Node.js scripts.

reskin