form-state-patterns
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION] (MEDIUM): The skill uses active shell execution to dynamically generate content based on the local environment.
- Evidence: Multiple instances of shell commands in
scripts/create-form.mdusing the!prefix, such as!grep -r "react-hook-form\|formik" package.jsonand!find . -name "*form*.tsx". - Risk: This capability allows the agent to inspect the directory structure and read file contents (like
package.json). If the shell execution environment is not properly sandboxed, it could be abused. - [DATA_EXPOSURE] (LOW): The skill explicitly reads the contents of
package.jsonto identify installed libraries. - Evidence:
grepcommands targetingpackage.jsonto extract library names. - [PROMPT_INJECTION] (LOW): User-provided arguments are interpolated directly into the output code template.
- Evidence:
export function $ARGUMENTS()and/** $ARGUMENTS Form */. - Risk: While the arguments are not passed to the shell commands in this specific skill, malicious input could be used to generate broken or deceptive React components.
Audit Metadata