migration-architect
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from user-supplied JSON files (schemas and migration specifications) to produce reports and runbooks.\n
- Ingestion points: Data enters the system via the
--input,--before, and--afterarguments inscripts/migration_planner.py,scripts/compatibility_checker.py, andscripts/rollback_generator.py.\n - Boundary markers: The processing scripts do not employ explicit boundary markers or delimiters when interpolating external data into generated prose or command templates.\n
- Capability inventory: The skill provides a suite of tools for generating shell commands (e.g.,
kubectl,aws cli,terraform) and SQL statements, which may be executed by an agent or user.\n - Sanitization: The scripts do not perform escaping or rigorous validation on input fields (such as table names or identifiers) before embedding them in output scripts, creating a risk for command or SQL injection if the inputs are crafted maliciously.\n- [DYNAMIC_EXECUTION]: The skill assembles executable scripts and command-line instructions at runtime using predefined templates.\n
- Evidence: Files
scripts/compatibility_checker.pyandscripts/rollback_generator.pycontain logic to construct SQL snippets (e.g.,ALTER TABLE {table_name} ADD COLUMN...) and shell commands based on parameters extracted from input files. These generated scripts are central to the skill's purpose but represent a dynamic execution surface that inherits the trust level of the input data.
Audit Metadata