hermes-in-buzz

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Downloads source code and release artifacts from the block/buzz GitHub repository and documentation from hermes-agent.nousresearch.com to guide the setup process.
  • [REMOTE_CODE_EXECUTION]: Clones external source code and performs local compilation using cargo build to create the necessary CLI tools and credential helpers.
  • [COMMAND_EXECUTION]: Executes system commands and subprocesses to configure platform-specific persistence (systemd, launchd, Task Scheduler) and manage Git operations. The skill uses a custom Python script to wrap these executions safely.
  • [CREDENTIALS_UNSAFE]: Interacts with sensitive .env files to manage Nostr private keys. The skill includes a dedicated credential manager that uses getpass to ensure secrets are provided via secure local input and never passed through command-line arguments.
  • [PROMPT_INJECTION]: Ingests data from GitHub release APIs and vendor documentation at hermes-agent.nousresearch.com. It lacks explicit boundary markers to delimit this external content. The skill has broad capabilities including file writing and shell execution through subprocesses. While no specific sanitization is applied to the remote instructions, the operations are scoped to a specific setup process and are required for maintaining compatibility with the remote service.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 04:25 PM
Security Audit — agent-trust-hub — hermes-in-buzz