The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform shell commands using the postplus CLI utility. These operations include listing connections, discovering available tools, and executing analytics reports. These commands represent the intended mechanism for the skill's measurement functionality and do not involve privileged system operations.
[INDIRECT_PROMPT_INJECTION]: By processing data from external Google Analytics and Search Console APIs, the skill has an indirect prompt injection attack surface. Malicious data within an analytics property could theoretically attempt to influence agent responses. (1) Ingestion points: External data is retrieved via postplus channels tools run and stored in local JSON files (e.g., ga4-report-result.json) before being read by the agent. (2) Boundary markers: The instructions provide cognitive boundaries by directing the agent to separate observed results from explanations and to report only what was actually measured. (3) Capability inventory: The skill's capabilities include shell execution of the postplus tool and writing results to the local filesystem via redirection. (4) Sanitization: The instructions focus on data accuracy and interpretation but do not include explicit technical sanitization for the content of the API responses.

measurement