# Modern Runtime headless continuation

## Current checkpoint

Slice J is closed. Slice K0-K3 are closed; K4 implementation is complete
through `776482da`, with the exact connected terminal gate active.

K3's production closeout commit is `3f340125`. Its automated predecessors are
`12b500d3`, `9569dadb`, and `b6547ff3`.

K4 commits after the telemetry/CPU checkpoint are:

- `97c174fb`: 1/5/10/30-root content sharing, same-GUID isolation, portal,
  reconnect, and 480,000-frame-per-root two-hour simulated endurance;
- `bd236ce5`: death during the endurance route plus five randomized
  30-session cancellation-edge runs;
- `776482da`: committed diagnostic-only 30-session resource ceilings,
  per-sample evaluation, named failures, and exact running/terminal ownership
  checks.

Windows and native Ubuntu Headless tests pass 67/67; the complete Release
suite passes 8,784 / 5 skips.

The Windows visible-terminal host and native Linux/WSL host both connected to
local ACE without loading presentation assemblies. Linux completed the
Aerlinthe/Caul/Rynthid/lifestone route, accepted external `teleallto`
destinations, remained visible to a retail observer, turned and ran smoothly,
and shut down through ACE-confirmed graceful logout. The user accepted the
observer movement gate on 2026-07-27.

## Important connected correction

`RuntimeLiveEntitySessionController` passes the exact accepted
`PositionTimestampDisposition` to the direct world projection.

- Normal `Apply` echoes update canonical state but do not reset the local
  headless movement controller.
- `ForcePosition` uses `BlipPosition`, preserves active movement semantics, and
  sends an immediate autonomous-position acknowledgement.
- A real teleport performs one full synchronization from
  `PrepareDestination`.

Do not reintroduce full local-player synchronization for every accepted
Position packet; it stops movement and causes retail observers to see repeated
run/snap-back rubberbanding.

Collision `[snap]` output is high volume and must remain behind the explicit
`PhysicsEngine.DiagnosticLog` sink.

## Automation seam

Graphical automation and headless policies use `IGameRuntimeView`,
`IGameRuntimeCommands`, `IRuntimeEventSource`, and the same
`RuntimeLocalPlayerFrameController`/outbound serializers. The route and
observer policies are diagnostic consumers, not a second bot world or
movement implementation.

The user requires optional direct single-session CLI launch with
`-user`/`-password` or the long-form aliases. Multi-session configuration
continues to use credential providers.

## Next work

Finish K4 from `docs/plans/2026-07-26-modern-runtime-slice-k.md`:

1. keep the exact `776482da` two-account Rynthid process running while the user
   observes it;
2. verify every periodic sample remains inside the committed envelope with no
   fault/reconnect/lease debt;
3. stop it gracefully and require a final `disposed` sample with two converged
   Runtime roots, zero entities/inventory/host leases, disposed shared
   content, and `withinCeilings=true`;
4. close Slice K and begin Slice L Linux graphical/platform work.

Final K3 automated counts: Runtime 412, Headless 47, App 3,722 / 3 skips,
complete Release 8,764 / 5 skips.

K3 connected-closeout rollback:

```text
git revert 3f3401257c83bdd9f557eb577ba08cf2e5ec9a06
```

K4 rollbacks, newest first:

```text
git revert 776482da8225d88a7c6128022d137d26f9266a87
git revert bd236ce553718cfb1248703f4a2b17b35733b11c
git revert 97c174fbb2d72fb94e2fe07448bb257989ed8941
git revert cb512fd0
```