google-sheets

Account setup could support AgentPMT-hosted Sheets tools, but the supplied code consists entirely of installing other skills and provides no Sheets-specific implementation or invocation. Its observable purpose is skill installation and onboarding, materially different from the advertised spreadsheet operations. The behavior of the installed skills cannot be determined from this chunk.

The related-skill installation command uses an unpinned npx skills executable and an unpinned skill repository reference. Although this is an installer rather than an MCP server, following the instruction could execute a compromised package version or install changed agent instructions; no actual compromise is demonstrated.

The onboarding installation example invokes an unversioned npm executable and retrieves skill content without an immutable revision. The setup context makes installation relevant, but it does not prevent upstream changes from introducing executable code or malicious agent instructions.

The account-setup installation command leaves both the installer version and downloaded skill revision unpinned. This is particularly sensitive because the installed instructions concern account connectivity and credentials, although the supplied file does not demonstrate credential theft.

The installation script repeats the unpinned onboarding command, exposing users who run it to mutable npm and repository content. This is the same supply-chain weakness as the earlier onboarding example, not a separate exploit mechanism.

The script installs the account/API setup skill through an unpinned executable and mutable repository reference. A compromised upstream dependency could influence subsequent account configuration or execute code through the installer; this repeats the earlier account-setup risk.

The reference section again recommends installing the onboarding skill with an unpinned npx command and repository revision. Its placement in documentation reduces immediacy but does not remove the supply-chain risk if an agent or user follows it.

The account-setup reference repeats an installation path whose executable and skill content can change independently of this file. Because the downloaded skill guides authentication setup, unauthorized upstream changes could affect a sensitive workflow.

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.