cli-forge-scaffold
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection via the interpolation of untrusted inputs into generated source code.
- Ingestion points: Data such as skill_name, description, author, and version are ingested from external inputs as specified in the Required Inputs table of instructions/new.md.
- Boundary markers: Placeholders use a double-curly brace token system but the generated outputs lack delimiters or sanitization logic when the tokens are placed inside Rust code blocks or doc comments in templates like main.rs.tpl.
- Capability inventory: The skill automatically executes cargo build, cargo clippy, and cargo test on the newly generated project directory as part of its verification workflow documented in SKILL.md.
- Sanitization: While the skill_name is validated against a specific kebab-case regex, other metadata fields such as description and author are interpolated directly into generated files without escaping characters that could potentially manipulate code execution during the verification phase.
Audit Metadata