Work item: WDD-003
Claim class: Competitive positioning evidence (not a product survey).
As of 2026-06-10, TeaAgent ships a release eval gate (scripts/run_release_eval_gate.py,
wired in .github/workflows/release.yml) with a conversational regression corpus.
Surveyed IDE-first agents (Cursor, Copilot, Windsurf) and hosted delegators (Codex cloud, Devin-style products) do not publish an equivalent blocking release eval gate tied to operator receipts and audit evidence. They optimize for latency, IDE integration, or hosted delegation — not governed release blocking.
The eval gate is a plausible public differentiator once S1–S3 truth work is complete and the gate stays green on release tags. This note does not claim superior model quality — only differentiated governance-at-release.
Next competitor landscape refresh (WS6-003 quarterly) or any competitor shipping a documented release-blocking eval gate.