zoho-mail
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use
npx @useclawlink/cli, which downloads and executes the ClawLink command-line interface from the NPM registry. - [COMMAND_EXECUTION]: The skill requires the execution of shell commands through the ClawLink CLI to manage Zoho Mail connections and actions.
- [CREDENTIALS_UNSAFE]: The skill manages authentication tokens locally by storing credentials in the
~/.clawlink/credentials.jsonfile following a browser-based OAuth flow. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The agent reads potentially untrusted email content from Zoho Mail via the
run zoho-mailaction. - Boundary markers: The skill explicitly includes a safety instruction: "confirm with the user before any write."
- Capability inventory: The skill can execute CLI commands and perform API-driven write operations (sending or updating emails) in the user's Zoho Mail account.
- Sanitization: The skill relies on human-in-the-loop confirmation rather than automated sanitization to mitigate risks from malicious data processed by the agent.
Audit Metadata