diff --git a/docs/reference/protocols/turn-envelope-v0.md b/docs/reference/protocols/turn-envelope-v0.md index 45f50f9323..5ca81d4967 100644 --- a/docs/reference/protocols/turn-envelope-v0.md +++ b/docs/reference/protocols/turn-envelope-v0.md @@ -623,8 +623,9 @@ diagnostic remain unchanged. This is an output qualification, not a full-cost SL The benchmark heartbeat worker verifies a historical `turn_envelope` request against the pinned CLI's v0 response before launching a model. Its read-only -preview and eligibility are recorded in the wake receipt. See the [runtime -guide](../../../benchmark/runtime/RUNTIME.md) for supported configurations. +preview and eligibility are recorded in the wake receipt. See the +[runtime guide](https://github.com/loopx-project/loopx/blob/main/benchmark/runtime/RUNTIME.md) +for supported configurations. Model cost, decision quality and long-run adoption require a separately pinned experiment. A capture carries its original read and recovery scope.