The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/doctor.py performs environment diagnostics by executing a Python probe using subprocess.run. This is used to verify the availability of necessary runtimes and libraries like uv, crawl4ai, and playwright within the local environment.
[EXTERNAL_DOWNLOADS]: As an SEO crawler, the skill intentionally fetches data from external sources. Scripts such as scripts/crawl_batch.py and scripts/seed_urls.py download HTML, robots.txt, and sitemap files from user-specified domains to perform their analysis.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it processes untrusted data from the web.
Ingestion points: Untrusted data enters via scripts/crawl_batch.py (site crawl) and scripts/seed_urls.py (sitemap discovery).
Boundary markers: The extracted metadata and content are saved in NDJSON format without delimiters to prevent the agent from interpreting data as instructions.
Capability inventory: The skill has the ability to make network requests, write files, and execute shell commands for environment checks.
Sanitization: Content is parsed using BeautifulSoup and HTMLParser, but there is no explicit sanitization to remove or neutralize potential instructions directed at the AI agent.

crawl4ai-seo