Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-24 · high
The test ran 1,802 Oracle scripts through Kiro, Gemini, and Copilot, then swapped specs between agents to see what broke.
Why now: The cross-agent numbers come from a paper posted to arXiv in August 2026, before most teams have audited spec handoffs between agent vendors.
Gemini's SQL syntax validity collapsed to 2.33% when it worked from a migration spec written by a different AI coding agent. Token F1 came out to 0.035, AST similarity to 0.015, in a test that swapped specs between agents to see what survived the handoff.
The numbers come from a paper posted to arXiv on Oracle-to-PostgreSQL migrations. Researchers ran 1,802 Oracle PL/SQL scripts, pulled from a corpus of 1,006 files, through three tools: Amazon Kiro, Gemini, and Copilot. Then they cross-tested, feeding a spec one agent wrote into a different agent's execution step. The worst pairing was Gemini running on a Kiro-authored spec.
Rewriting the foreign spec into Gemini's own format before execution closed most of the gap.
The gap matters for anyone chaining different vendors' coding agents on the same migration job, since this test covered only a few of many possible agent pairings. It's a narrow slice of the broader spec-interchange problem, but the collapse was steep enough to notice.
Each link below shares sources, entities, or timing with this story.
Copilot uses Claude / Shared entities / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (Copilot uses Claude); both cover Copilot, Gemini; reported by the same outlet (arxiv.org).
Oracle supports MCP / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Oracle supports MCP); both cover Copilot, Kiro; overlapping topics (agent, between, copilot).
Copilot deprecates Gemini / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Copilot deprecates Gemini); both cover Copilot, Gemini; overlapping topics (copilot, directly, gemini).
Copilot uses Claude / Shared entity: Gemini / Same source domain / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Copilot uses Claude); both cover Gemini; reported by the same outlet (arxiv.org).
Oracle supports MCP / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Oracle supports MCP); both cover Copilot, Gemini; overlapping topics (agent, design).
AI Agent Studio built by Oracle / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (AI Agent Studio built by Oracle); both cover Gemini, Oracle; overlapping topics (agent, gemini).
Cursor benchmarked against Kiro / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Cursor benchmarked against Kiro); both cover Copilot, Kiro; overlapping topics (agent, design).
Oracle supports MCP / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Oracle supports MCP); both cover Copilot, Gemini; overlapping topics (between, copilot).