The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing the gog CLI tool to perform Apps Script operations. It emphasizes safety by recommending the use of --readonly, --dry-run, and --no-input flags to prevent accidental or unauthorized mutations of Google data.
[INDIRECT_PROMPT_INJECTION]: The skill addresses the risk of processing untrusted data from external Apps Script projects.
Ingestion points: External content is retrieved using the content and pull commands as documented in SKILL.md.
Boundary markers: The documentation recommends using the --wrap-untrusted flag to properly scope external content for the agent.
Capability inventory: The skill leverages the gog CLI for project management and execution.
Sanitization: The --wrap-untrusted flag is highlighted as a best practice to ensure that external content is handled safely within the agent's context.

gog-appscript