The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the scrapfly CLI via go install from the vendor's GitHub repository. Additionally, the exit-peer install command downloads necessary binaries and certificates from the vendor's infrastructure to establish an mTLS tunnel for proxying.
[COMMAND_EXECUTION]: The skill defines a wide array of shell commands using the scrapfly CLI to perform web scraping, screenshots, and browser control. This includes complex operations like starting persistent browser daemons and running autonomous agent loops.
[DYNAMIC_EXECUTION]: The tool provides an eval command to execute arbitrary JavaScript within a remote cloud browser session. In its agent mode, the skill allows an LLM to dynamically determine and execute these tool calls based on natural language tasks.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to fetch and process untrusted content from the public web for LLM consumption.
Ingestion points: External data enters the agent context via scrapfly scrape, scrapfly crawl, and scrapfly extract commands as documented in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the agent to treat scraped content as untrusted.
Capability inventory: The skill allows the agent to perform network requests, write to the local filesystem using output flags, and interact with web applications through clicks, form fills, and script evaluation.
Sanitization: While the CLI supports structured output formats like markdown and clean_html which provide basic structural filtering, it does not explicitly sanitize the text content for malicious prompt instructions.

scrapfly-cli