mise

Fail

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation section recommends the command curl https://mise.run | sh, which downloads and executes a remote script directly in the shell without integrity verification.
  • [EXTERNAL_DOWNLOADS]: The skill downloads content from https://mise.run and references external documentation/registries at https://mise.jdx.dev and https://direnv.net.
  • [COMMAND_EXECUTION]: The skill uses eval to hook Mise and Direnv into the shell environment (e.g., eval "$(mise activate zsh)"), which involves executing output generated by external binaries.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its reliance on repository-local configuration.
  • Ingestion points: Processes configuration from .mise.toml and .envrc files within a project directory.
  • Boundary markers: None; there are no instructions to validate the contents of these files before allowing them.
  • Capability inventory: The mise install and direnv allow commands can trigger binary installation and environment variable execution based on file content.
  • Sanitization: The skill relies on manual trust and allow commands, which serve as security boundaries but may be bypassed if an agent automates setup tasks in untrusted or malicious repositories.
Recommendations
  • HIGH: Downloads and executes remote code from: https://mise.run - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Apr 7, 2026, 03:50 PM