The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from the web, which could contain instructions designed to manipulate the agent's behavior.
Ingestion points: Web content retrieved via browse snapshot, browse get text, and browse get html (SKILL.md, REFERENCE.md).
Boundary markers: The instructions do not define clear boundaries or include directives to ignore instructions embedded in the extracted web content.
Capability inventory: The skill utilizes the Bash tool to execute browse commands, which include eval for JavaScript execution, screenshot for file writing, and open for network operations (SKILL.md, REFERENCE.md).
Sanitization: No evidence of sanitization or filtering of the content extracted from external websites.
[EXTERNAL_DOWNLOADS]: Installs the browse CLI tool from the official NPM registry to provide the necessary browser automation functionality.
[COMMAND_EXECUTION]: Uses the Bash tool to run CLI commands for automating browser interactions, including navigation, element interaction, and session management.
[DATA_EXFILTRATION]: Handles the BROWSERBASE_API_KEY environment variable for authenticated remote browser sessions and supports local network traffic capture via the browse network on command.

browser