Skip to content

Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure

TypeScript

docs(llm-proxy): state that the eight stream slots are per proxy process

Alfredo Lopezd967d906efdd
diff --git a/docs/architecture/codex-subscription-transport-contract.md b/docs/architecture/codex-subscription-transport-contract.md
index 606342e1d..5bc29bcb5 100644
--- a/docs/architecture/codex-subscription-transport-contract.md
+++ b/docs/architecture/codex-subscription-transport-contract.md
@@ -183,6 +183,9 @@ must still be `authorized`, and the consuming `UPDATE` matches only
| maxRetriesPerAttempt | 1 |
| executionTicketTtlMs | 60000 |
+`maxConcurrentStreams` is 8 slots per proxy process, shared by every Host;
+per-Host fairness is tracked in #767.
+
Tool definitions have no independent count ceiling in the Evenfire request
contract. The entire serialized request, including all definitions, remains
bounded by `maxRequestBodyBytes`. Every definition still undergoes
diff --git a/docs/architecture/grok-subscription-transport-contract.md b/docs/architecture/grok-subscription-transport-contract.md
index 39ed6af4d..c8e1ebfde 100644
--- a/docs/architecture/grok-subscription-transport-contract.md
+++ b/docs/architecture/grok-subscription-transport-contract.md
@@ -152,6 +152,9 @@ The limit values live only in the table above, which
same suite pins the fixture to the contract `LIMITS` and the proxy
`STREAM_LIMITS`.
+`maxConcurrentStreams` is 8 slots per proxy process, shared by every Host;
+per-Host fairness is tracked in #767.
+
All three enforcement points read this module — the control-api authorizer,
`grok-llm-proxy` and the Host — so a deployment that mixes versions rejects
requests that fall between the old and the new bounds. Which code the caller