The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from potentially untrusted external sources, which could be exploited to influence agent behavior through hidden instructions.
Ingestion points: SKILL.md instructs the agent to gather questions from the "owner's support inbox, contact-form submissions, DMs, and reviews."
Boundary markers: The instructions lack specific boundary markers or instructions to ignore embedded commands within these untrusted inputs.
Capability inventory: The agent has the capability to write new files to the filesystem (to create the FAQ page) and execute the scripts/check-links.mjs script.
Sanitization: There is no mention of sanitizing or escaping the data retrieved from support messages before it is processed or used for content generation.
[DYNAMIC_EXECUTION]: The script scripts/lib/browser.mjs performs dynamic loading of modules using paths computed at runtime.
Evidence: The importPlaywright function utilizes createRequire().resolve() to find the installation path of playwright or playwright-core and then executes import() on the resolved file URL.
[COMMAND_EXECUTION]: The skill requires the execution of local Node.js scripts to perform its verification steps.
Evidence: The Verify section of SKILL.md explicitly directs the agent to run node scripts/check-links.mjs --url <site> --render to check link integrity.
[EXTERNAL_DOWNLOADS]: The included utility script performs automated network requests to external third-party domains discovered at runtime.
Evidence: In scripts/check-links.mjs, the checkExternal function uses fetchWithRetry to connect to external URLs found within the target project's HTML to verify their status codes.
[DATA_EXFILTRATION]: The skill's operational flow involves accessing and reading sensitive user communications which presents an exposure and exfiltration risk.
Evidence: SKILL.md directs the agent to inspect "support inbox, contact-form submissions, DMs, and reviews" to identify real-world user questions. This involves the handling of private user data that could be exfiltrated if the agent is directed to an external URL.

faq-readiness