Cross‑Language Support for Post‑Emit Execution in TypeSpec SDK Generation

[lirenhe]

Background

As part of ongoing efforts to automate TypeSpec to SDK, service teams (Foundry SDK) identified a gap in the current SDK generation workflow: there is no standardized way to run post‑generation logic after SDK code is emitted.

This is discussed in Foundry Spec ↔ SDK auto‑sync discussion, where people raised scenarios requiring post‑emit fixes or adjustments (temporary SDK workarounds, transitional fixes, or one‑time corrections) that cannot cleanly live inside the emitter logic today. Service teams also expressed the need for flexibility for customization. We proposed the approach of post‑emit execution step, ideally supported directly by emitters, with temporary pipeline‑level solutions as a short‑term fallback.

Similar issue was also discussed Azure/azure-sdk-tools#10366 where multiple design decision were discussed:

• tsp-client should not become a generic post‑processing runner.
• Post‑generation logic should be owned by emitters, not centrally orchestrated.

This epic tracks the work required to achieve language parity by introducing or aligning post‑emit execution support in each language emitter, which is triggered by real scenarios raised in Foundry requirements.

We have this effort for GO today to handle some special request for KeyVault: Azure/autorest.go#1636

[lirenhe]

Please hold this effort now until we get more clarification from our architects.