The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the browse CLI to perform actions such as navigating pages, inspecting browser state, and managing remote infrastructure. This includes powerful commands like browse eval for JavaScript execution and browse stop --force for process management.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the browse global npm package and enables the discovery and installation of site-specific automation skills from the Browserbase catalog (browse.sh). These resources are managed by the vendor and provide verified strategies for interacting with specific domains.
[REMOTE_CODE_EXECUTION]: The skill manages the development and invocation of Browserbase Functions, which are serverless scripts executed on the vendor's cloud infrastructure. Additionally, the browse eval command allows for the execution of arbitrary JavaScript within the context of a browser session.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks because it extracts and processes content from external websites via snapshots, text reads, and HTML capture. This content is then provided to the agent, creating a surface for embedded instructions to influence agent behavior.
Ingestion points: Data enters the context through browse snapshot, browse get text, browse get markdown, and browse cloud fetch (SKILL.md).
Boundary markers: The skill does not define specific delimiters or warnings to isolate untrusted web content from the agent's core instructions.
Capability inventory: The agent has access to the Bash tool, file upload capabilities via browse upload, and JavaScript execution via browse eval (SKILL.md).
Sanitization: No explicit sanitization or filtering of the retrieved web content is mentioned in the skill instructions.

browse