The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill automates web browser interactions, which involves processing data from untrusted external websites. \n
Ingestion points: Untrusted data enters the agent's context through agent-browser snapshot (accessibility trees) and agent-browser get (text/HTML) commands as documented in SKILL.md. \n
Boundary markers: There are no explicit instructions or delimiters mentioned to ensure the agent ignores malicious instructions embedded in the content of the pages it visits. \n
Capability inventory: The skill allows for navigation, element interaction (clicking, typing), and session state management. \n
Sanitization: The skill does not specify any sanitization or filtering of data retrieved from the web browser. \n
[EXTERNAL_DOWNLOADS]: The documentation includes instructions to use npx @claude-flow/cli for memory storage and search functions. \n
This results in the dynamic download and execution of the package from the NPM registry. \n
[COMMAND_EXECUTION]: The skill relies on the execution of CLI tools, specifically agent-browser and npx, to perform its core functions.

browser