The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's SKILL.md file instructs the user to install the bdata CLI by piping a shell script from https://cli.brightdata.com/install.sh directly to bash. This script is hosted on the official infrastructure of the skill's author, Bright Data, making it a standard and expected deployment method for the tool.
[COMMAND_EXECUTION]: The skill extensively utilizes the shell to execute bdata CLI commands for web scraping and data pipeline tasks. These commands are dynamically populated with external URLs and search terms.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from various external websites, creating a surface for indirect prompt injection.
Ingestion points: External web data from competitor sites, G2, Capterra, and LinkedIn retrieved via bdata calls in SKILL.md.
Boundary markers: The instructions in SKILL.md provide structured templates and guidelines to 'Separate facts from analysis' and 'Cite every data point,' but there are no explicit delimiters to segregate untrusted data from the system prompt.
Capability inventory: Execution of shell commands via the bdata tool is utilized across all modules in SKILL.md.
Sanitization: There is no evidence of sanitization or filtering of the scraped content within the provided skill instructions.
[EXTERNAL_DOWNLOADS]: The skill performs numerous network requests to external services like Crunchbase, LinkedIn, Amazon, and review aggregators to collect market intelligence, utilizing the vendor's web scraping infrastructure.

competitive-intel