use-ravion
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the Ravion CLI installation script from the vendor's official GitHub releases page and fetches a Model Context Protocol (MCP) server configuration from the vendor's documentation site.
- [REMOTE_CODE_EXECUTION]: The skill executes the downloaded Ravion CLI installer via the shell (
sh ravion-install.sh) and runs a remote MCP server setup usingnpxfrom a vendor-controlled URL. - [COMMAND_EXECUTION]: The skill makes extensive use of shell tools to run the
ravionandawsCLIs for cloud provisioning,gitfor repository management, andbrewfor package installation. - [PRIVILEGE_ESCALATION]: The skill initiates the creation of a high-privilege IAM role using AWS CloudFormation (
CAPABILITY_NAMED_IAM). This is a legitimate requirement for the platform's functionality, and the instructions mandate that the agent verifies the AWS account identity and obtains explicit user approval before execution. - [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill analyzes repository manifest files, such as
package.json,pyproject.toml, andDockerfile, to automatically detect and generate infrastructure configurations. - Boundary markers: The skill does not specify the use of delimiters or specific "ignore instruction" markers when processing the content of these manifest files.
- Capability inventory: The agent possesses capabilities for shell command execution, AWS infrastructure management, and project file modification.
- Sanitization: There is no mention of sanitization or validation of the content extracted from repository files before it is used to influence the generated infrastructure code.
Audit Metadata