Skip to content
Closed
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -43,15 +43,26 @@ import { servers as evalsTracingConfig } from '../../evals_tracing/stateful/clas
* alert-analysis, alert-triage, workflow-authoring, entity-analytics (v1 path),
* detection-rule-edit (base), find-security-ml-jobs.
*
* `server.maxPayload` is raised from the Scout default (1.6 MB) because the
* judge's `/internal/inference/prompt` requests carry the full agent trajectory
* and exceed it on long multi-step examples (observed 413 on
* gemini-3.1-flash-lite, 2026-08-20). The inherited arg is filtered out and
* re-set here so the value is deterministic regardless of CLI arg ordering.
*
* Usage:
* node scripts/scout start-server --arch stateful --domain classic --serverConfigSet evals_security_persona_matrix
*/
const MAX_PAYLOAD_BYTES = 50 * 1024 * 1024; // 50 MB

export const servers: ScoutServerConfig = {
...evalsTracingConfig,
kbnTestServer: {
...evalsTracingConfig.kbnTestServer,
serverArgs: [
...evalsTracingConfig.kbnTestServer.serverArgs,
...evalsTracingConfig.kbnTestServer.serverArgs.filter(
(arg) => !arg.startsWith('--server.maxPayload=')
),
`--server.maxPayload=${MAX_PAYLOAD_BYTES}`,
'--uiSettings.overrides.agentBuilder:experimentalFeatures=true',
`--xpack.securitySolution.enableExperimental=${JSON.stringify([
'automaticTroubleshootingSkill',
Expand Down
Loading