mise
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project configuration files, which creates a surface for indirect prompt injection.
- Ingestion points: Ingests mise.toml, mise.local.toml, and .tool-versions, and follows _.file or _.source directives to load environment variables and scripts.
- Boundary markers: None are present to verify the source or integrity of the ingested files.
- Capability inventory: Includes powerful execution capabilities through mise exec and mise run, which can run arbitrary shell commands.
- Sanitization: Lacks specific input sanitization for configuration content, although output redaction is supported for sensitive environment variables.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading tools from external package registries and repositories.
- Evidence: References github, npm, and pipx backends for tool acquisition.
- Context: These downloads target well-known services and repositories as part of the tool's core function.
- [COMMAND_EXECUTION]: Shell commands are utilized for managing the toolchain and environment.
- Evidence: Instructions include using mise doctor, mise config, and mise install to manage project state.
Audit Metadata