The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill instructions require the user to disable the agent's security sandbox by starting it with the '--sandbox danger-full-access' flag. This removes safety boundaries and allows the skill to operate with the user's full system permissions.
[DYNAMIC_EXECUTION]: The skill relies on a persistent 'js_repl' environment and uses 'page.evaluate' and 'electronApp.evaluate' to execute arbitrary JavaScript within the context of the browser or Electron application.
[COMMAND_EXECUTION]: Setup and execution involve shell commands to initialize projects with 'npm init', install packages, and launch local Electron binaries via the Playwright library.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection when processing data from external web pages or local applications. (1) Ingestion points: Untrusted content is ingested through browser navigation in 'SKILL.md' and interactions with Electron windows. (2) Boundary markers: There are no explicit markers or instructions to ignore embedded commands within the processed data. (3) Capability inventory: The skill has the ability to execute shell commands and perform dynamic code evaluation. (4) Sanitization: No sanitization or content validation is performed on the data retrieved from the browser or Electron contexts.
[EXTERNAL_DOWNLOADS]: The skill downloads and installs the 'playwright' and 'electron' packages from official registries and fetches browser binaries using 'npx playwright install'.

playwright-interactive