The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from external websites.
Ingestion points: Web content is fetched in scripts/basic_crawl.py (line 103) and scripts/deep_crawl.py (line 173) via the crawl4ai library.
Boundary markers: The scripts do not implement clear boundary markers or instructions to prevent the agent from following commands embedded in the crawled content.
Capability inventory: The skill can perform network requests and write files to the local filesystem.
Sanitization: There is no evidence of sanitization or filtering of the Markdown content extracted from web pages.
[COMMAND_EXECUTION]: The skill uses uv run to execute local Python scripts. The scripts/deep_crawl.py script generates output file paths using URL path segments (self.output_dir / f"{path}.md"). Because it only performs simple stripping of slashes, it may be vulnerable to a path traversal attack if a malicious starting URL with directory traversal sequences is provided.
[EXTERNAL_DOWNLOADS]: The skill specifies crawl4ai as a dependency, which is a legitimate third-party library for web crawling and is installed from the standard Python package registry.

skill-browser-crawl