encore-migrate
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an existing source codebase to generate migration plans and source code for a new Encore project.
- Ingestion points: Phase 1, Step 2 involves reading the entire source codebase at a user-provided path.
- Boundary markers: The instructions do not specify any delimiters or safety warnings to ignore instructions embedded within the source files (e.g., in code comments or strings).
- Capability inventory: The skill has the capability to write files to the file system (via the
migration-plan/directory) and delegate code generation to other skills (encore-service,encore-api, etc.), which in turn write code. It also performs local HTTP requests for validation. - Sanitization: No explicit sanitization or validation of the source code content is mentioned before it is used to populate the migration plan or passed to other skills.
- [COMMAND_EXECUTION]: The skill performs HTTP requests to local URLs during the validation phase to compare responses between the source and target systems. While the instructions state to ask for user confirmation before making calls with side effects, this interaction with local network services constitutes a capability that could be misused if influenced by malicious input.
Audit Metadata