raw-app
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to execute specific CLI commands using the
wmilltool (e.g.,wmill app new,wmill generate-metadata,wmill app lint) to automate the creation and maintenance of application files. - [PROMPT_INJECTION]: The instructions include directives for the agent to perform CLI actions directly rather than instructing the user to do so manually. While designed to streamline automation, this instruction can be interpreted as a technique to reduce process transparency.
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an environment for creating applications that ingest and process external data.
- Ingestion points: The agent collects application metadata (summary, path), database configuration (datatable, schema), and table names from user input and project files.
- Boundary markers: The agent is instructed to use structured questions and confirm setup values with the user before proceeding.
- Capability inventory: The skill enables the agent to write frontend and backend source code, manage configuration files, execute shell commands via the CLI, and perform database operations.
- Sanitization: The instructions explicitly mandate the use of parameterized queries and tagged templates in both TypeScript and Python to prevent SQL injection vulnerabilities.
Audit Metadata