brev-cli

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Shell tool to run brev CLI and ssh commands locally for instance lifecycle management, hardware discovery, and port-forwarding operations.
  • [REMOTE_CODE_EXECUTION]: The skill supports executing commands on remote Brev instances using the brev exec utility for environment verification and application deployment.
  • [DATA_EXFILTRATION]: The skill uses brev copy for file transfers and includes specific instructions to prevent the accidental upload of sensitive .env files and local secrets.
  • [PROMPT_INJECTION]: The skill processes JSON output from brev ls and brev search, which constitutes an indirect prompt injection surface.
  • Ingestion points: Data enters the agent context from shell command outputs in SKILL.md.
  • Boundary markers: No specific delimiters isolate the tool output.
  • Capability inventory: The skill has access to Shell, Filesystem, ssh, and brev tools.
  • Sanitization: No explicit validation or filtering is performed on the tool's JSON output before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 06:56 AM
Security Audit — agent-trust-hub — brev-cli