@sema-agent/server 1.186.0 → 1.188.0

package/dist/http/server.js

@@ -181,6 +181,14 @@ export function createHttpServer(deps) {
     let uncountedBillableInflight = 0;
     if (deps.drainState)
         deps.drainState.inflight = () => new Set([...inflightRuns.keys(), ...steerableRuns.keys()]).size + uncountedBillableInflight + (deps.leaderEndpoint?.inflight?.() ?? 0);
+    let lastBillableActivityAt = Date.now();
+    const noteActivity = (url) => {
+        if (url === "/health" || url.startsWith("/metrics"))
+            return;
+        lastBillableActivityAt = Date.now();
+    };
+    if (deps.drainState)
+        deps.drainState.lastActivityAt = () => lastBillableActivityAt;
     if (deps.hookWakeBus) {
         deps.hookWakeBus.deliver = async (sessionId, text) => {
             try {
@@ -249,6 +257,7 @@ export function createHttpServer(deps) {
         const startedAt = Date.now();
         const method = req.method ?? "GET";
         const url = (req.url ?? "/").split("?")[0];
+        noteActivity(url);
         let logged = false;
         const record = (viaClose) => {
             if (logged)
@@ -674,6 +683,7 @@ export function createHttpServer(deps) {
                 }, 15_000);
                 let ranLive = false;
                 let resp;
+                uncountedBillableInflight++;
                 try {
                     resp = await idemCache.run(idemKey, async () => {
                         ranLive = true;
@@ -850,9 +860,12 @@ export function createHttpServer(deps) {
                             const withElicit = deps.elicitation
                                 ? () => deps.elicitation.runWithContext({ taskId: askTaskId, owner: askOwner, emit: emitAsk, abortSignal: ac.signal }, streamBody)
                                 : streamBody;
-                            await (deps.question
+                            const withQuestion = () => deps.question
                                 ? deps.question.runWithContext({ taskId: askTaskId, owner: askOwner, emit: emitAsk, abortSignal: ac.signal }, withElicit)
-                                : withElicit());
+                                : withElicit();
+                            await (deps.sendUserFile
+                                ? deps.sendUserFile.runWithContext({ taskId: askTaskId, emit: async (f) => void (await emitAsk(f)) }, withQuestion)
+                                : withQuestion());
                             if (durableTaskId && deps.runStore) {
                                 if (finalResult?.status === "suspended" && deps.checkpointStore)
                                     await deps.runStore.setSuspended(durableTaskId);
@@ -893,6 +906,7 @@ export function createHttpServer(deps) {
                         throw e;
                 }
                 finally {
+                    uncountedBillableInflight--;
                     clearInterval(hb);
                 }
                 if (!ranLive && resp && !res.writableEnded) {
@@ -1033,7 +1047,7 @@ export function createHttpServer(deps) {
                     rootTaskId: taskId,
                     ...fleetRunLabels(prepared.spec.objective),
                 });
-                void runInBackground(deps.runner, { ...prepared.spec, sessionId }, runStore, taskId, deps.metrics, prepared.auth?.principal, prepared.verify?.maxRounds, prepared.cascade ? cascadeConfig(deps.config.cascadeLadder, prepared.spec.maxCostUsd) : undefined, deps.instrumentDegenerate, deps.planCacheProbe, deps.config.traceThinking, inflightRuns, preemptableRuns, steerableRuns, deps.modelUsage, deps.elicitation, gatedPrincipal(req, deps.config) ?? null, captureTurnAnchor, fleetPub, captureUserMessageAnchor, deps.question, deps.workflowCompletionInbox, deps.subagentSteerRegistry, (m, x) => deps.logger?.info?.(m, x));
+                void runInBackground(deps.runner, { ...prepared.spec, sessionId }, runStore, taskId, deps.metrics, prepared.auth?.principal, prepared.verify?.maxRounds, prepared.cascade ? cascadeConfig(deps.config.cascadeLadder, prepared.spec.maxCostUsd) : undefined, deps.instrumentDegenerate, deps.planCacheProbe, deps.config.traceThinking, inflightRuns, preemptableRuns, steerableRuns, deps.modelUsage, deps.elicitation, gatedPrincipal(req, deps.config) ?? null, captureTurnAnchor, fleetPub, captureUserMessageAnchor, deps.question, deps.workflowCompletionInbox, deps.subagentSteerRegistry, (m, x) => deps.logger?.info?.(m, x), deps.sendUserFile);
                 return { status: 202, body: { taskId, sessionId, status: "running" } };
             }, (r) => r.status === 202);
             sendJson(res, resp.status, resp.body);