himalaya
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute various
himalayaCLI commands for email management, including listing envelopes and reading messages. - [COMMAND_EXECUTION]: It uses shell commands like
pass showorsecurity find-generic-passwordto retrieve authentication credentials securely from the system's password store. - [EXTERNAL_DOWNLOADS]: The skill metadata defines an installation process using the Homebrew package manager (
brew install himalaya) to provide the necessary binary. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests untrusted data from external email servers.
- Ingestion points: Email bodies and headers are retrieved via
himalaya message readandhimalaya envelope list. - Boundary markers: There are no explicit instructions or delimiters used to separate untrusted email content from the agent's instructions during processing.
- Capability inventory: The agent has the ability to send emails (
himalaya template send), download attachments to local directories (himalaya attachment download), and execute shell commands for credential retrieval. - Sanitization: No sanitization is performed on email content before it is processed or included in reply templates.
- [SAFE]: The skill prioritizes secure authentication methods, documenting the use of system keyrings and password managers over raw text storage, and uses well-known installation sources.
Audit Metadata