The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill allows the agent to ingest content from Google Docs via the gws docs documents get command, which represents an attack surface for indirect prompt injection where instructions embedded in a document could override agent behavior. The skill lacks boundary markers or sanitization logic to distinguish between data and instructions. The agent is provided with write capabilities (batchUpdate, create, and the +write helper), which could be leveraged if a malicious document influences the agent's output.
[COMMAND_EXECUTION]: The skill's primary function is to execute shell commands via the gws utility. While this is intentional, the agent's ability to run these commands to modify cloud resources based on potentially untrusted input requires human oversight to ensure that commands generated from processed document data are safe and intended.

gws-docs