The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the gws command-line tool to interact with Google Sheets API resources such as spreadsheets, values, and developerMetadata. It provides instructions in SKILL.md for the agent to execute shell commands to perform these operations.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection as it retrieves data from external Google Sheets which is then processed by the agent.
Ingestion points: Untrusted data enters the agent context through spreadsheet reading methods (e.g., spreadsheets.get, spreadsheets.values.get) documented in SKILL.md.
Boundary markers: The skill does not define specific delimiters or warnings within the prompt instructions to ensure the agent disregards instructions found within the spreadsheet data.
Capability inventory: The skill allows both reading from and writing to the API via the gws binary, creating a read-process-write cycle.
Sanitization: There is no evidence of sanitization, filtering, or validation of the retrieved spreadsheet content before it is presented to the agent in the provided instructions.

gws-sheets