The Agent Skills Directory

[AWS Authentication Integration]: The skill utilizes standard AWS credential resolution patterns, including environment variables like AWS_ACCESS_KEY_ID and the aws CLI. This is the expected and secure method for authenticating with AWS services when running cloud-hosted browsers.
[Restricted Command Execution]: The skill's capabilities are scoped to the agent-browser utility through the allowed-tools configuration. This limits the agent's actions to the specific browser automation tasks intended by the author.
[Web Content Processing Surface]: As a tool designed to interact with the live web, the skill processes external data which is a standard aspect of browser automation.
Ingestion points: Untrusted data is ingested from external URLs via agent-browser open and agent-browser snapshot operations described in SKILL.md.
Boundary markers: No specific delimiters or "ignore instructions" markers are explicitly defined for the data retrieved from browser sessions.
Capability inventory: The skill uses agent-browser to perform interactions like clicking, filling forms, and taking screenshots as noted in SKILL.md.
Sanitization: The skill relies on the standard security boundaries provided by the cloud browser environment for content handling.

agentcore