Salvaged from #12565 (@Oxidane-bot) — just the callback-ownership portion. The /status totals half of that PR was already fixed on main in 7abc9ce via #17158.

Summary

Stale runs could fire a fresher run's post-delivery callback because the generation-ownership check was silently bypassed.

Root cause

_process_message_background in gateway/platforms/base.py snapshotted callback_generation at the top of the task:

interrupt_event = self._active_sessions.get(session_key) or asyncio.Event()
self._active_sessions[session_key] = interrupt_event
callback_generation = getattr(interrupt_event, "_hermes_run_generation", None)

But _hermes_run_generation is only set on the event by GatewayRunner._bind_adapter_run_generation during _handle_message_with_agent — which runs inside the await self._message_handler(event) below. The early snapshot always captured None.

That None then flowed into pop_post_delivery_callback(..., generation=None) in the finally block. Inside pop, generation=None with a tuple-registered entry bypasses the entry_generation != generation check, pops, and fires the callback regardless of which run owns it.

Fix

Move the snapshot into the finally block, after the handler has run and _hermes_run_generation has been bound.

Validation

New regression test: test_post_delivery_callback_generation_snapshot_happens_after_bind

• Simulates a stale handler at generation=1 and a fresher callback registered at generation=2
• Pre-fix: snapshot=None → pop fires the generation=2 callback under generation=1's ownership (fired == ['newer'])
• Post-fix: snapshot=1 → pop skips the mismatched entry (fired == [])

Verified: test FAILS on current main, PASSES with this fix. Reverted the base.py change locally to confirm the test actually catches the bug.

Credit

Authored by @Oxidane-bot (from #12565), with a Co-authored-by trailer. Also adds them to scripts/release.py AUTHOR_MAP.