The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The scripts/run.mjs file uses the AsyncFunction constructor to execute JavaScript code passed as strings via command-line arguments or standard input. This allows the agent to generate and run arbitrary code on the host system within the automation environment.
[PROMPT_INJECTION]: The skill processes untrusted data from the 1C web client, creating a surface for indirect prompt injection where malicious content in the browser could influence the agent's actions.
Ingestion points: Data enters the agent context through readForm, readTable, and readSpreadsheet functions in scripts/dom.mjs.
Boundary markers: The instructions do not define boundary markers to separate retrieved data from agent instructions.
Capability inventory: The skill has access to the Bash tool, file system access via Read and Write, and the ability to execute scripts via scripts/run.mjs.
Sanitization: No validation or sanitization is performed on the data retrieved from the web client.
[EXTERNAL_DOWNLOADS]: The installation process (scripts/package.json) downloads the Playwright library and browser binaries. Documentation in recording.md also guides users to download ffmpeg from external builds for video features. These are well-known technology tools and services.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform environment setup and manage the execution of automation scripts via Node.js.

web-test