Repeatability
Medium
Converting a raw SQL query to a parameterized Dapper or EF query follows a repeatable pattern, but each query has unique joins, business logic, and edge cases that require contextual interpretation. The 40-query scope and 'high-traffic' selection step introduce non-uniform judgment.
Ambiguity Tolerance
Medium
Success criteria are partially defined (5 queries, parameterization, unit tests) but 'query optimization' and 'high-traffic' selection are vague without profiling data or business context. The agent cannot self-verify correctness against a live database.
Data & Tool Availability
Medium
The agent needs full access to the C# codebase, database schema, and ideally query execution plans or traffic logs — none of which are guaranteed to be provided. Without the schema and existing query context, the agent is guessing at table relationships and data types.
Error Cost
High
Incorrect ORM mappings or missed parameterization can introduce SQL injection vulnerabilities or silent data corruption in production. Errors in high-traffic queries have outsized impact and may not surface until runtime.
Human Judgment Required
High
Selecting which 5 queries are truly high-traffic requires profiling data and business priority knowledge. Optimization decisions (indexing hints, eager vs. lazy loading, N+1 avoidance) require understanding of real usage patterns that an agent cannot infer from code alone.