stream-chain
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by design as it is built to process untrusted data through sequential agent steps.\n
- Ingestion points: The skill is designed to ingest data from source code directories (e.g., code smells identification), CSV files (user data extraction), and API responses (data transformation) as described in
SKILL.md.\n - Boundary markers: There are no explicit instructions or examples of using delimiters or boundary markers to segregate untrusted data from instructions within the workflow.\n
- Capability inventory: The skill enables significant capabilities including code refactoring, test generation, and automated data processing, which could be exploited if malicious instructions are embedded in the ingested data (documented in
SKILL.md).\n - Sanitization: There is no mention of sanitization, validation, or escaping of ingested data before it is processed by the agent or passed to subsequent steps in the chain.
Audit Metadata