The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Clones source code from a third-party GitHub repository (Google-Health-API) that is not included in the list of trusted vendors or organizations. \n- [COMMAND_EXECUTION]: Instructions require the agent or user to compile the tool from source using go build and execute the resulting binary on the local system. \n- [PRIVILEGE_ESCALATION]: Recommends the use of sudo to move the compiled binary into a system-wide executable path (/usr/local/bin/), which grants the installation process administrative rights. \n- [DATA_EXFILTRATION]: Accesses and manages highly sensitive personal health metrics and OAuth authentication tokens stored in ~/.config/ghealth/. The skill also provides an export command (ghealth auth export) that could be used to retrieve credentials in plain text. \n- [PROMPT_INJECTION]: The skill ingests untrusted health data (such as activity notes and exercise descriptions) from the external Google Health API. This content is processed and presented to the agent without explicit sanitization or boundary markers, creating a potential surface for indirect prompt injection. Ingestion points: Data retrieved via ghealth data list and get commands. Boundary markers: None present. Capability inventory: Local command execution and file system access. Sanitization: No sanitization or validation of external content described.

google-health-cli