The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/lib/browser.mjs utilizes dynamic imports with paths computed at runtime. Specifically, it uses createRequire().resolve() to locate playwright or playwright-core and then imports the resulting file URL. While intended for optional dependency management, dynamic loading from computed paths is a significant execution vector.
[COMMAND_EXECUTION]: The skill launches a headless browser process via Playwright's chromium.launch method in scripts/lib/browser.mjs. It iterates through several common system paths to locate a Chromium or Chrome executable and supports a user-provided binary path via the --chrome-path argument, which could be exploited to execute arbitrary binaries if the argument is manipulated.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by fetching and parsing HTML from external URLs, creating a vulnerability surface for indirect prompt injection.
Ingestion points: scripts/lib/pages.mjs fetches HTML content from external origins using the fetch API via the loadUrlSite function.
Boundary markers: The skill does not implement explicit boundary markers or delimiters when processing the fetched HTML content.
Capability inventory: The skill has the capability to make network requests (fetch), execute shell commands/subprocesses (browser launch), and read from the local filesystem.
Sanitization: The custom HTML parser in scripts/lib/html.mjs decodes standard entities but lacks specific sanitization or filtering to prevent embedded instructions from influencing the agent's behavior.

link-integrity