The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill interacts with external, untrusted web pages and prototypes. It implements a robust defense-in-depth strategy by instructing the agent to treat all webpage content as untrusted and prohibiting the revelation of secrets or scope expansion based on page-supplied data.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill follows security best practices by requiring all identity and session evidence to be sanitized and non-reversible. It explicitly forbids the persistence of PII, such as email addresses, cookies, or authentication tokens.
[COMMAND_EXECUTION]: The included Python scripts serve as local utility tools for route resolution and schema validation. They use only standard Python libraries and do not perform arbitrary shell execution or download remote code.
[DYNAMIC_CONTEXT_INJECTION]: Analysis confirms the absence of dynamic execution patterns (such as !command syntax) that could lead to silent command execution at skill load time.

ops-browser