The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires fetching the workspace-admin-pp-cli tool from an external third-party GitHub repository (github.com/mvanhorn/printing-press-library) and the NPM registry. These sources are not included in the global trusted vendor list, representing an unverifiable dependency for a tool that acquires significant administrative control over a Google Workspace domain.
[REMOTE_CODE_EXECUTION]: The installation instructions utilize npx -y and go install to fetch and execute code from an external repository. The -y flag in the npx command allows for silent, automated execution of remote packages without user confirmation, which represents a remote code execution vector during the skill's setup phase.
[DATA_EXFILTRATION]: The CLI tool includes an output delivery system supporting a webhook:<url> sink. This feature allows the results of any command—which contain highly sensitive Google Workspace metadata, PII, and security posture information—to be automatically POSTed to an arbitrary external URL, creating a direct path for data exfiltration if the agent is misdirected.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted organizational data (including Drive file metadata, Gmail message headers, and audit logs) into a local SQLite store for processing. This data is then used by the agent to drive high-privilege administrative actions.
Ingestion points: File, message, and user metadata from Google Workspace Directory, Drive, Gmail, Reports, and Alert Center APIs.
Boundary markers: The skill uses the --agent flag to produce structured JSON output, but does not provide explicit instruction delimiters to isolate untrusted data from the agent's prompt context.
Capability inventory: The skill possesses powerful administrative capabilities including workflow offboard (suspend user, sign out, revoke tokens, move OUs), device-wipe, delete-filter, and send-message.
Sanitization: There is no mention of sanitization, validation, or filtering of the synced organizational data before it is processed by the agent.

pp-workspace-admin