kotlin-specialist
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts on user-provided code and project descriptions to generate implementations and run analysis tools. It does not define explicit boundary markers or delimiters for this untrusted input, creating a potential surface for indirect prompt injection through user-supplied code snippets.\n
- Ingestion points: User-provided data models, code snippets, and architectural strategies (SKILL.md).\n
- Boundary markers: None present; the skill lacks instructions to isolate user content.\n
- Capability inventory: Generates Kotlin code and invokes static analysis tools like
detektandktlint(SKILL.md).\n - Sanitization: No explicit sanitization or escaping of user input is specified before it enters the agent's context.\n- [METADATA_POISONING]: The frontmatter metadata contains a URL in the
authorfield (https://github.com/Jeffallan) instead of a plain name. While likely intended for attribution, metadata fields containing links are a known vector for redirection.
Audit Metadata