mirth-to-ballerina
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided Mirth Connect
channel.xmlfiles to generate Ballerina source code, creating a surface for potential injection attacks.\n - Ingestion points: The skill accepts a Mirth channel XML from the user as input to drive the migration process (SKILL.md).\n
- Boundary markers: The instructions do not define explicit delimiters (like XML tags or markdown blocks) to isolate the user's input XML from the agent's internal instructional context.\n
- Capability inventory: The skill generates code with extensive capabilities, including SQL database access (ballerina/sql), network communication (http:Client, tcp:Listener, health.clients.hl7, health.clients.fhir), and file system operations (ballerina/file) (references/connector-activity-mappings.md).\n
- Sanitization: While the skill promotes structural parsing into typed records, it lacks explicit logic or instructions for the agent to sanitize or ignore instructions that may be embedded within the XML fields, which could potentially manipulate the generated output.
Audit Metadata