gmail-access

Warn

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires several shell commands for environment setup, building the project with Bun, and running the Gmail CLI binary.
  • [EXTERNAL_DOWNLOADS]: The 'bun install' command downloads Node.js dependencies from the public NPM registry during the setup phase.
  • [REMOTE_CODE_EXECUTION]: The skill triggers a build and execution process for a custom CLI from a local directory path; however, the source code for this CLI is not provided in the analyzed files, making its behavior unverifiable.
  • [DATA_EXFILTRATION]: The tool is designed to read sensitive Gmail messages, search inboxes, and access 1Password items, exposing this sensitive data to the agent's context.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface. Ingestion points: Untrusted data enters the context from Gmail emails via list, search, and read operations. Boundary markers: None are specified to prevent the agent from following instructions found within emails. Capability inventory: The agent has access to powerful tools including Bash and Write. Sanitization: No sanitization or filtering of email content is performed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 3, 2026, 10:05 PM