wasm-spa-autofix-react-imports
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by ingesting and acting upon potentially untrusted data provided in inputs to generate file modifications.
- Ingestion points: The skill processes
bundlerLogsandfileContentsas primary inputs to determine necessary code changes (SKILL.md). - Boundary markers: There are no explicit instructions or delimiters defined to differentiate between legitimate log data and potential malicious instructions embedded within those logs or files.
- Capability inventory: The skill is designed to generate
patchesthat modify source code, which is a high-impact capability if the generation process is subverted. - Sanitization: While the skill includes instructions to "Validate patches syntactically," it lacks sanitization logic to prevent an attacker from influencing the patch generation via adversarial content placed in build logs or source files.
Audit Metadata