Skip to content

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

TypeScript

fix: classify proxy 413 as context length and derive the body limit (#731)

Alfredo Lopez2b04bd6a077c
diff --git a/codex-llm-proxy/src/config.ts b/codex-llm-proxy/src/config.ts
index 2baa3094d..50a43fc79 100644
--- a/codex-llm-proxy/src/config.ts
+++ b/codex-llm-proxy/src/config.ts
@@ -1,3 +1,5 @@
+import { DEFAULT_MAX_BODY_BYTES } from './requestLimits.js'
+
export type CodexLlmProxyConfig = {
runtimePort: number
adminPort: number
@@ -63,7 +65,7 @@ export function loadConfig(env: NodeJS.ProcessEnv = process.env): CodexLlmProxyC
maxBodyBytes: requiredPositiveInt(
'CODEX_LLM_PROXY_MAX_BODY_BYTES',
env.CODEX_LLM_PROXY_MAX_BODY_BYTES,
- 1_048_576
+ DEFAULT_MAX_BODY_BYTES
),
maxStreamDurationMs: requiredPositiveInt(
'CODEX_LLM_PROXY_MAX_STREAM_DURATION_MS',
diff --git a/codex-llm-proxy/src/requestLimits.ts b/codex-llm-proxy/src/requestLimits.ts
index 550e48cde..4c06401f8 100644
--- a/codex-llm-proxy/src/requestLimits.ts
+++ b/codex-llm-proxy/src/requestLimits.ts
@@ -1,5 +1,18 @@
import { LIMITS as CONTRACT_LIMITS } from '@clerum/llm-provider-attempt-contract'
+/**
+ * #731 — room for the runtime envelope around the contract-capped `request`:
+ * the execution ticket (a few KB by its claim bounds), the request hash, the
+ * deadline and the JSON keys. 16 KiB is several times that.
+ */
+export const ENVELOPE_ALLOWANCE_BYTES = 16 * 1024
+
+/**
+ * The default body limit, derived from the contract so the proxy never refuses,
+ * as a 413, a request the contract accepts.
+ */
+export const DEFAULT_MAX_BODY_BYTES = CONTRACT_LIMITS.maxRequestBodyBytes + ENVELOPE_ALLOWANCE_BYTES
+
export const STREAM_LIMITS = {
maxConcurrentStreams: 8,
maxQueuedRequests: 16,
diff --git a/deploy/base/control-plane/codex-llm-proxy.yaml b/deploy/base/control-plane/codex-llm-proxy.yaml
index 71e3abb29..42f6b7379 100644
--- a/deploy/base/control-plane/codex-llm-proxy.yaml
+++ b/deploy/base/control-plane/codex-llm-proxy.yaml
@@ -18,7 +18,6 @@ data:
CODEX_LLM_PROXY_RUNTIME_PORT: "8080"
CODEX_LLM_PROXY_ADMIN_PORT: "8081"
CODEX_LLM_PROXY_PROBE_PORT: "9090"
- CODEX_LLM_PROXY_MAX_BODY_BYTES: "1048576"
CODEX_LLM_PROXY_MAX_STREAM_DURATION_MS: "300000"
CODEX_LLM_PROXY_MAX_DEADLINE_MS: "300000"
CODEX_LLM_PROXY_JWT_ISSUER: control-api
diff --git a/deploy/base/control-plane/grok-llm-proxy.yaml b/deploy/base/control-plane/grok-llm-proxy.yaml
index 0ef0e570c..831653870 100644
--- a/deploy/base/control-plane/grok-llm-proxy.yaml
+++ b/deploy/base/control-plane/grok-llm-proxy.yaml
@@ -18,7 +18,6 @@ data:
GROK_LLM_PROXY_RUNTIME_PORT: "8080"
GROK_LLM_PROXY_ADMIN_PORT: "8081"
GROK_LLM_PROXY_PROBE_PORT: "9090"
- GROK_LLM_PROXY_MAX_BODY_BYTES: "1048576"
GROK_LLM_PROXY_MAX_STREAM_DURATION_MS: "300000"
GROK_LLM_PROXY_MAX_DEADLINE_MS: "300000"
GROK_LLM_PROXY_JWT_ISSUER: control-api
diff --git a/grok-llm-proxy/src/config.ts b/grok-llm-proxy/src/config.ts
index 92f8d3996..c266547c8 100644
--- a/grok-llm-proxy/src/config.ts
+++ b/grok-llm-proxy/src/config.ts
@@ -1,3 +1,5 @@
+import { DEFAULT_MAX_BODY_BYTES } from './requestLimits.js'
+
export type GrokLlmProxyConfig = {
runtimePort: number
adminPort: number
@@ -75,7 +77,7 @@ export function loadConfig(env: NodeJS.ProcessEnv = process.env): GrokLlmProxyCo
maxBodyBytes: requiredPositiveInt(
'GROK_LLM_PROXY_MAX_BODY_BYTES',
env.GROK_LLM_PROXY_MAX_BODY_BYTES,
- 1_048_576
+ DEFAULT_MAX_BODY_BYTES
),
maxStreamDurationMs: requiredPositiveInt(
'GROK_LLM_PROXY_MAX_STREAM_DURATION_MS',
diff --git a/grok-llm-proxy/src/requestLimits.ts b/grok-llm-proxy/src/requestLimits.ts
index ba22dd3b7..f0cc6d4bd 100644
--- a/grok-llm-proxy/src/requestLimits.ts
+++ b/grok-llm-proxy/src/requestLimits.ts
@@ -1,5 +1,18 @@
import { LIMITS as CONTRACT_LIMITS } from '@clerum/grok-provider-attempt-contract'
+/**
+ * #731 — room for the runtime envelope around the contract-capped `request`:
+ * the execution ticket (a few KB by its claim bounds), the request hash, the
+ * deadline and the JSON keys. 16 KiB is several times that.
+ */
+export const ENVELOPE_ALLOWANCE_BYTES = 16 * 1024
+
+/**
+ * The default body limit, derived from the contract so the proxy never refuses,
+ * as a 413, a request the contract accepts.
+ */
+export const DEFAULT_MAX_BODY_BYTES = CONTRACT_LIMITS.maxRequestBodyBytes + ENVELOPE_ALLOWANCE_BYTES
+
export const STREAM_LIMITS = {
maxConcurrentStreams: 8,
maxQueuedRequests: 16,
diff --git a/mcp-host/package-lock.json b/mcp-host/package-lock.json
index e9ff67041..640fbe173 100644
--- a/mcp-host/package-lock.json
+++ b/mcp-host/package-lock.json
@@ -1,12 +1,12 @@
{
"name": "clerum-mcp-host",
- "version": "0.1.390",
+ "version": "0.1.391",
"lockfileVersion": 3,
"requires": true,
"packages": {
"": {
"name": "clerum-mcp-host",
- "version": "0.1.390",
+ "version": "0.1.391",
"license": "SEE LICENSE IN LICENSE",
"dependencies": {
"@anthropic-ai/sdk": "^0.32.0",
diff --git a/mcp-host/package.json b/mcp-host/package.json
index 727b96eab..86cf2d996 100644
--- a/mcp-host/package.json
+++ b/mcp-host/package.json
@@ -1,6 +1,6 @@
{
"name": "clerum-mcp-host",
- "version": "0.1.390",
+ "version": "0.1.391",
"description": "MCP Host - reads Host CRD and provides LLM access via OpenAI/Claude",
"main": "dist/main.js",
"engines": {
diff --git a/mcp-host/src/llm/codexSubscription.ts b/mcp-host/src/llm/codexSubscription.ts
index 1d52c4cdb..01c588543 100644
--- a/mcp-host/src/llm/codexSubscription.ts
+++ b/mcp-host/src/llm/codexSubscription.ts
@@ -232,7 +232,9 @@ export class CodexSubscriptionProvider implements SingleTurnProvider {
...(providerDispatched !== undefined ? { providerDispatched } : {}),
}
}
- if (code === 'request_limit_exceeded') {
+ // `payload_too_large` is the proxy's 413 for an envelope over its body
+ // limit: the same size refusal of this conversation, one hop later (#731).
+ if (code === 'request_limit_exceeded' || code === 'payload_too_large') {
return {
code: LlmErrorCode.ContextLengthExceeded,
retryable: false,
diff --git a/mcp-host/src/llm/grokSubscription.ts b/mcp-host/src/llm/grokSubscription.ts
index 84749c958..ef286fbfd 100644
--- a/mcp-host/src/llm/grokSubscription.ts
+++ b/mcp-host/src/llm/grokSubscription.ts
@@ -252,7 +252,13 @@ export class GrokSubscriptionProvider implements SingleTurnProvider {
// transport refused a tool call whose arguments overran its size budget.
// Treating it as an outage would retry the identical oversized call and,
// being failover-eligible, spend a second provider on it.
- if (code === 'request_limit_exceeded' || code === 'tool_call_arguments_exceeded') {
+ // `payload_too_large` is the proxy's 413 for an envelope over its body
+ // limit: the same size refusal of this conversation, one hop later (#731).
+ if (
+ code === 'request_limit_exceeded' ||
+ code === 'tool_call_arguments_exceeded' ||
+ code === 'payload_too_large'
+ ) {
return {
code: LlmErrorCode.ContextLengthExceeded,
retryable: false,