The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to scrape external web content, which presents a surface for indirect prompt injection attacks.
Ingestion points: Data enters the agent context through Fetcher.get, DynamicFetcher.fetch, StealthyFetcher.fetch, and the Spider class in SKILL.md.
Boundary markers: The instructions do not define boundary markers or instruct the agent to ignore embedded instructions within the scraped content.
Capability inventory: The skill allows network requests and file output via the CLI (scrapling extract). Scripts using the library could potentially perform more sensitive operations with the retrieved data.
Sanitization: There is no evidence of content sanitization or filtering to prevent the agent from following malicious instructions embedded in scraped HTML or Markdown.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the scrapling package and its browser dependencies.
Evidence: The installation process uses pip install scrapling[all] followed by scrapling install. The latter is a CLI command that typically downloads browser binaries (such as Chromium or Firefox) to the local system.
[COMMAND_EXECUTION]: The skill utilizes a CLI tool (scrapling) and Python scripts to perform complex browser automation and network operations. While these are documented and intended features, they represent a significant capability set that should be monitored.

scrapling