Repeatability
High
SQL dialect translation follows well-documented, consistent rules (AUTOINCREMENT → SERIAL/IDENTITY, strftime → date_trunc/to_char, etc.). The same transformation logic applies to each of the 8 queries with minimal case-by-case judgment.
Ambiguity Tolerance
High
Success is objectively measurable: the rewritten queries must execute on PostgreSQL and produce equivalent output to the originals on SQLite. Unit tests make the pass/fail criterion explicit and machine-verifiable.
Data & Tool Availability
Medium
The agent needs access to the codebase (the 8 queries), the Django schema, and ideally both a SQLite and PostgreSQL instance to run equivalence tests. If the repo and a test DB environment are provided, this is fully executable; without them, the agent can only produce untested code.
Error Cost
Medium
Incorrect query rewrites could silently return wrong data in production, which is a real risk. However, the task is scoped to a dev/migration context with unit tests as a safety net, and changes are code-reviewable before deployment — making errors detectable and reversible.
Human Judgment Required
Low
The translation rules are well-established and the scope is narrow. A human should do a final review for subtle semantic edge cases, but the core work — identifying syntax differences and rewriting queries — does not require intuition or domain taste.