@sema-agent/core 3.0.0 → 4.0.0

package/dist/agents/subagent.js

@@ -11,6 +11,7 @@ import { SUBAGENT_PROMPT } from "../prompts/default.js";
 import { hasSessionFork } from "../core/session.js";
 import { isDurablePause } from "./suspend-guard.js";
 import { delimitUntrusted, inlineUntrusted } from "../core/untrusted-text.js";
+import { createSafeNotifier } from "../core/safe-notify.js";
 import { boundedRedactedSummary } from "../core/untrusted-egress.js";
 import { uuidv7 } from "../internal/harness.js";
 import { addWorktree } from "../core/git-worktree-env.js";
@@ -18,6 +19,17 @@ import { shellQuote } from "../tools/fs/search.js";
 import { BG_AGENT_REAP_STOP_ERROR } from "../core/task-registry.js";
 import { extractErrorCode } from "../brain/errors.js";
 import { FORK_DEFAULT_MAX_TURNS, SESSION_BG_DEFAULT_TIMEOUT_SEC, RETAIN_DEFAULT_TTL_MS, RETAIN_DEFAULT_MAX, RUNNING_AGENT_OBSERVE_EVERY_BEATS } from "../config/defaults.js";
+function rollupDelegatedCost(stats, nested) {
+    if (stats.costMicroUsd === undefined)
+        return undefined;
+    if (nested !== undefined && nested.costMicroUsd === undefined)
+        return undefined;
+    return stats.costMicroUsd + (nested?.costMicroUsd ?? 0);
+}
+function delegatedCostField(stats) {
+    const total = rollupDelegatedCost(stats, stats.nested);
+    return total !== undefined ? { costMicroUsd: total } : {};
+}
 export { FORK_DEFAULT_MAX_TURNS, SESSION_BG_DEFAULT_TIMEOUT_SEC, RETAIN_DEFAULT_TTL_MS, RETAIN_DEFAULT_MAX };
 import { SUBAGENT_RESUME_CAP, SubagentRetainLedger, getOrCreateSessionRetainLedger, ensureSessionReapHook, createResumePrompt, } from "./retain-ledger.js";
 import { recordRosterSpawn } from "./roster-store.js";
@@ -264,6 +276,7 @@ function createSteerHandle(stream, parentToolCallId, agentName, settled, retain)
     };
 }
 export function createSubagentResume(deps) {
+    const notifier = createSafeNotifier();
     const resume = async (content) => {
         const ledger = deps.ledger;
         if (!ledger) {
@@ -403,6 +416,7 @@ export function createSubagentResume(deps) {
                                     taskId: deps.taskId,
                                     sessionScoped: deps.sessionScoped === true,
                                     ...(deps.rowAgentType !== undefined ? { agentType: deps.rowAgentType } : {}),
+                                    sessionId: entry.childSessionId,
                                     transcriptId: entry.childSessionId,
                                     ...(deps.parentToolCallId !== undefined ? { parentToolCallId: deps.parentToolCallId } : {}),
                                     progressTaskId: e.taskId,
@@ -442,11 +456,7 @@ export function createSubagentResume(deps) {
             settleResolve = r;
         });
         if (deps.sink) {
-            try {
-                deps.sink(createSteerHandle(stream, deps.parentToolCallId, entry.agentName, settled, { resume, childSessionId: entry.childSessionId }));
-            }
-            catch {
-            }
+            notifier.notify(() => deps.sink?.(createSteerHandle(stream, deps.parentToolCallId, entry.agentName, settled, { resume, childSessionId: entry.childSessionId })), "subagent.resume.sink");
         }
         void (async () => {
             try {
@@ -493,8 +503,9 @@ export function createSubagentResume(deps) {
                         seq: entry.cycleSeq,
                         ...(stoppedByRevive !== undefined ? { stoppedBy: stoppedByRevive } : {}),
                         summary: reviveTerminalSummary,
+                        ...resumeResidual(),
                         resumable: status !== "killed",
-                        usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: child.stats.costMicroUsd },
+                        usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                         ...(completionIdRevive !== undefined ? { completionId: completionIdRevive } : {}),
                     });
                 const resumeFrame = {
@@ -512,7 +523,7 @@ export function createSubagentResume(deps) {
                     ...(status === "killed" && child.result ? { partial: true } : {}),
                     ...resumeResidual(),
                     resumable: status !== "killed",
-                    usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: child.stats.costMicroUsd },
+                    usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                     ...(completionIdRevive !== undefined ? { completionId: completionIdRevive } : {}),
                 };
                 const gateOwner = ownsTerminalFacesRevive && status === "completed" && deps.notify !== undefined && deps.registry !== undefined && ledger.get(deps.parentToolCallId) === entry
@@ -777,6 +788,7 @@ export function agentWhenToUseText(def, lean = true) {
 function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
     const maxDepth = opts.maxDepth ?? DEFAULT_SUBAGENT_MAX_DEPTH;
     const toolName = opts.name ?? DEFAULT_SUBAGENT_TOOL_NAME;
+    const notifier = createSafeNotifier();
     const deploymentNames = new Set((opts.agents ?? []).map((a) => a.name));
     const builtins = opts.builtinAgents === false ? [] : builtinAgentDefinitions(toolName).filter((d) => !deploymentNames.has(d.name));
     const available = [...(opts.agents ?? []), ...builtins].filter((a) => !excluded.has(a.name));
@@ -1153,8 +1165,9 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                 : perCallModel ?? (def ? (def.model ?? ctx.model ?? opts.model) : opts.model);
             const childSystemPrompt = def?.systemPrompt ?? opts.systemPrompt;
             const childDefaultPersona = childSystemPrompt === undefined && !wantsFork ? SUBAGENT_PROMPT : undefined;
-            const childLimits = def?.maxTurns !== undefined
-                ? { ...opts.limits, maxTurns: def.maxTurns }
+            const defMaxTurns = typeof def?.maxTurns === "number" && Number.isFinite(def.maxTurns) && def.maxTurns > 0 ? def.maxTurns : undefined;
+            const childLimits = defMaxTurns !== undefined
+                ? { ...opts.limits, maxTurns: defMaxTurns }
                 : wantsFork && opts.limits?.maxTurns === undefined
                     ? { ...opts.limits, maxTurns: FORK_DEFAULT_MAX_TURNS }
                     : opts.limits;
@@ -1188,14 +1201,10 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                 else if (observerFailure === undefined) {
                     observerFailure = msg;
                 }
-                try {
-                    opts.onObserverError?.(err, {
-                        ...(def !== undefined ? { observedAgent: def.name } : {}),
-                        ...(observerArm ? { observerAgent: observerArm.observerDefinition.name } : {}),
-                    });
-                }
-                catch {
-                }
+                notifier.notify(() => opts.onObserverError?.(err, {
+                    ...(def !== undefined ? { observedAgent: def.name } : {}),
+                    ...(observerArm ? { observerAgent: observerArm.observerDefinition.name } : {}),
+                }), "subagent.onObserverError");
             };
             const observedSteerRef = {};
             let pendingTrigger = [];
@@ -1239,7 +1248,10 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                     ...(ctx.alwaysLoadTools !== undefined ? { alwaysLoadTools: [...ctx.alwaysLoadTools] } : {}),
                     ...(ctx.promptProfile !== undefined ? { promptProfile: ctx.promptProfile } : {}),
                     enableBlockedReport: false,
-                    limits: { maxTurns: observerDef.maxTurns ?? 8, timeoutSec: 120 },
+                    limits: {
+                        maxTurns: typeof observerDef.maxTurns === "number" && Number.isFinite(observerDef.maxTurns) && observerDef.maxTurns > 0 ? observerDef.maxTurns : 8,
+                        timeoutSec: 120,
+                    },
                 });
                 const runObserverLeg = async (spec) => {
                     if (!isObserverTaskId(sid))
@@ -1683,21 +1695,17 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                             resumable: false,
                             ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
                         });
-                        try {
-                            notify?.({
-                                task_id: taskId,
-                                task_type: "background_agent",
-                                toolUseId: ctx.toolCallId,
-                                status: "killed",
-                                stoppedBy: stoppedByReap,
-                                summary: summaryReap,
-                                sessionId: forkedId,
-                                resumable: false,
-                                ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
-                            }, { priority: "later" });
-                        }
-                        catch {
-                        }
+                        notifier.notify(() => notify?.({
+                            task_id: taskId,
+                            task_type: "background_agent",
+                            toolUseId: ctx.toolCallId,
+                            status: "killed",
+                            stoppedBy: stoppedByReap,
+                            summary: summaryReap,
+                            sessionId: forkedId,
+                            resumable: false,
+                            ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
+                        }, { priority: "later" }), "subagent.reapTerminalNotify");
                     });
                     if (agentName !== undefined) {
                         recordRosterSpawn(ctx.roster, { name: agentName, agentId: taskId, sessionId: forkedId, toolUseId: ctx.toolCallId, owner: bgOwner, scope: bgScope, ...((typeof childModel === "string" ? resolveModelDisplayLabel(childModel) : childModel?.id) !== undefined ? { model: typeof childModel === "string" ? childModel : childModel?.id } : {}), ...((ctx.rootSessionId ?? ctx.sessionId) !== undefined ? { rootSessionId: ctx.rootSessionId ?? ctx.sessionId } : {}), ...(sessionScopedBg ? { sessionScoped: true } : {}), createdAt: Date.now() });
@@ -1770,6 +1778,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                                         kind: "tick",
                                         taskId,
                                         sessionScoped: sessionScopedBg === true,
+                                        sessionId: forkedId,
                                         transcriptId: forkedId,
                                         parentToolCallId: ctx.toolCallId,
                                         progressTaskId: e.taskId,
@@ -1805,7 +1814,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                                 tokens: child.stats.tokens + (nsb?.tokens ?? 0),
                                 turns: child.stats.turns + (nsb?.turns ?? 0),
                                 tasks: 1 + (nsb?.tasks ?? 0),
-                                costMicroUsd: (child.stats.costMicroUsd ?? 0) + (nsb?.costMicroUsd ?? 0),
+                                costMicroUsd: rollupDelegatedCost(child.stats, nsb),
                             });
                         }
                         catch {
@@ -1857,7 +1866,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                                 summary: forkTerminalSummary,
                                 ...residualFork,
                                 resumable: resumableFork,
-                                usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: (child.stats.costMicroUsd ?? 0) + (child.stats.nested?.costMicroUsd ?? 0) },
+                                usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                                 ...(completionIdFork !== undefined ? { completionId: completionIdFork } : {}),
                             });
                         try {
@@ -1876,7 +1885,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                                     ...(settledBg === "killed" && child.result ? { partial: true } : {}),
                                     ...residualFork,
                                     resumable: resumableFork,
-                                    usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: (child.stats.costMicroUsd ?? 0) + (child.stats.nested?.costMicroUsd ?? 0) },
+                                    usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                                     ...(completionIdFork !== undefined ? { completionId: completionIdFork } : {}),
                                 }, { priority: "later" });
                         }
@@ -1980,7 +1989,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                     tokens: forkChild.stats.tokens + (nsf?.tokens ?? 0),
                     turns: forkChild.stats.turns + (nsf?.turns ?? 0),
                     tasks: 1 + (nsf?.tasks ?? 0),
-                    costMicroUsd: (forkChild.stats.costMicroUsd ?? 0) + (nsf?.costMicroUsd ?? 0),
+                    costMicroUsd: rollupDelegatedCost(forkChild.stats, nsf),
                 });
                 if (isDurablePause(forkChild.status)) {
                     return {
@@ -2131,21 +2140,17 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                         resumable: false,
                         ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
                     });
-                    try {
-                        notify?.({
-                            task_id: taskId,
-                            task_type: "background_agent",
-                            toolUseId: ctx.toolCallId,
-                            status: "killed",
-                            stoppedBy: stoppedByReap,
-                            summary: summaryReap,
-                            sessionId: bgChildSessionId,
-                            resumable: false,
-                            ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
-                        }, { priority: "later" });
-                    }
-                    catch {
-                    }
+                    notifier.notify(() => notify?.({
+                        task_id: taskId,
+                        task_type: "background_agent",
+                        toolUseId: ctx.toolCallId,
+                        status: "killed",
+                        stoppedBy: stoppedByReap,
+                        summary: summaryReap,
+                        sessionId: bgChildSessionId,
+                        resumable: false,
+                        ...(completionIdReap !== undefined ? { completionId: completionIdReap } : {}),
+                    }, { priority: "later" }), "subagent.reapTerminalNotify");
                 });
                 const bgDurableProbe = bg.registry.durableAgentRowProbe(taskId);
                 const bgRowConfirmed = async () => (bgDurableProbe ? await bgDurableProbe().catch(() => false) : false);
@@ -2249,6 +2254,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                                     kind: "tick",
                                     taskId,
                                     sessionScoped: sessionScopedBg === true,
+                                    sessionId: bgChildSessionId,
                                     transcriptId: bgChildSessionId,
                                     parentToolCallId: ctx.toolCallId,
                                     progressTaskId: e.taskId,
@@ -2405,7 +2411,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                             tokens: child.stats.tokens + (ns2?.tokens ?? 0),
                             turns: child.stats.turns + (ns2?.turns ?? 0),
                             tasks: 1 + (ns2?.tasks ?? 0),
-                            costMicroUsd: (child.stats.costMicroUsd ?? 0) + (ns2?.costMicroUsd ?? 0),
+                            costMicroUsd: rollupDelegatedCost(child.stats, ns2),
                         });
                     }
                     catch {
@@ -2515,7 +2521,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                             summary: bgTerminalSummary,
                             ...residual,
                             resumable: resumableBg,
-                            usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: (child.stats.costMicroUsd ?? 0) + (child.stats.nested?.costMicroUsd ?? 0) },
+                            usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                             ...(completionIdBg !== undefined ? { completionId: completionIdBg } : {}),
                         });
                     const completionFrame = {
@@ -2533,7 +2539,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                         ...(settled === "killed" && child.result ? { partial: true } : {}),
                         ...residual,
                         resumable: resumableBg,
-                        usage: { tokens: child.stats.tokens, turns: child.stats.turns, costMicroUsd: (child.stats.costMicroUsd ?? 0) + (child.stats.nested?.costMicroUsd ?? 0) },
+                        usage: { tokens: child.stats.tokens, turns: child.stats.turns, ...delegatedCostField(child.stats) },
                         ...(completionIdBg !== undefined ? { completionId: completionIdBg } : {}),
                     };
                     const gateOwner = ownsTerminalFaces && settled === "completed" && notify !== undefined ? bgRetain?.childSessionId : undefined;
@@ -2685,14 +2691,10 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                         sink: ctx.onSubagentSpawn,
                         ...(opts.background ? { registry: opts.background.registry } : {}),
                     });
-                    try {
-                        ctx.onSubagentSpawn(createSteerHandle(stream, ctx.toolCallId, childAgentName, settled, {
-                            resume,
-                            ...(retainEntry ? { childSessionId: retainEntry.childSessionId } : {}),
-                        }));
-                    }
-                    catch {
-                    }
+                    notifier.notify(() => ctx.onSubagentSpawn?.(createSteerHandle(stream, ctx.toolCallId, childAgentName, settled, {
+                        resume,
+                        ...(retainEntry ? { childSessionId: retainEntry.childSessionId } : {}),
+                    })), "subagent.onSubagentSpawn");
                 }
                 try {
                     for await (const ev of stream)
@@ -2737,7 +2739,7 @@ function createSubagentToolNode(opts, depth, excluded, extraToolsBudget) {
                 tokens: child.stats.tokens + (ns?.tokens ?? 0),
                 turns: child.stats.turns + (ns?.turns ?? 0),
                 tasks: 1 + (ns?.tasks ?? 0),
-                costMicroUsd: (child.stats.costMicroUsd ?? 0) + (ns?.costMicroUsd ?? 0),
+                costMicroUsd: rollupDelegatedCost(child.stats, ns),
             });
             if (isDurablePause(child.status)) {
                 if (retainEntry)

package/dist/agents/teacher.d.ts

@@ -53,15 +53,7 @@ export interface TeacherRunResult extends TaskResult {
         tokens: number;
         tasks: number;
         costMicroUsd?: number;
-        humanReview?: {
-            count: number;
-            totalWaitMs: number;
-            gates: Array<{
-                kind: string;
-                waitMs: number;
-                decision?: string;
-            }>;
-        };
+        humanReview?: NonNullable<TaskResult["stats"]["humanReview"]>;
     };
 }
 export declare function parseTeacherAdvice(text: string): TeacherAdvice;

package/dist/agents/teacher.js

@@ -2,6 +2,7 @@ import { randomUUID } from "node:crypto";
 import { mapNestedSuspend, isDurablePause } from "./suspend-guard.js";
 import { isDefineToolProduct, stampDefineToolBrand } from "../core/tools.js";
 import { delimitUntrusted, inlineUntrusted } from "../core/untrusted-text.js";
+import { buildCumulativeStats } from "./cumulative-stats.js";
 export const TEACHER_PROMPT = `You are an expert advisor to a less-capable "student" agent that got stuck.
 You receive the task and the student's recent failed attempts. Your job is to help the student RECOVER,
 not to do the work for it.
@@ -84,6 +85,7 @@ async function runTeacherCore(runner, studentSpec, teacher) {
     let studentTokens = 0;
     let studentTurns = 0;
     let studentCostMicro = 0;
+    let anyStudentLegUnpriced = false;
     let studentPrompt = 0;
     let studentTotalInput = 0;
     let studentCached = 0;
@@ -202,6 +204,8 @@ async function runTeacherCore(runner, studentSpec, teacher) {
         studentTokens += s.tokens;
         studentTurns += s.turns;
         studentCostMicro += s.costMicroUsd ?? 0;
+        if (s.costMicroUsd === undefined)
+            anyStudentLegUnpriced = true;
         studentPrompt += s.promptTokens ?? 0;
         studentTotalInput += s.totalInputTokens ?? 0;
         studentCached += s.cachedTokens ?? 0;
@@ -442,34 +446,27 @@ async function runTeacherCore(runner, studentSpec, teacher) {
     function finalize(result) {
         return {
             ...result,
-            stats: {
-                ...result.stats,
+            stats: buildCumulativeStats(result.stats, {
                 tokens: studentTokens,
                 turns: studentTurns,
                 costMicroUsd: studentCostMicro || undefined,
-                promptTokens: studentPrompt || undefined,
-                totalInputTokens: studentTotalInput || undefined,
-                cachedTokens: studentCached || undefined,
-                outputTokens: studentOutput || undefined,
-                cacheWriteTokens: studentCacheWrite || undefined,
-                cacheWriteTokensLong: studentCacheWriteLong || undefined,
-                cacheHitRate: studentTotalInput > 0 ? Math.min(1, studentCached / studentTotalInput) : undefined,
-                nested: studentNestedTokens > 0 || studentNestedTasks > 0
-                    ? {
-                        tokens: studentNestedTokens,
-                        turns: studentNestedTurns,
-                        tasks: studentNestedTasks,
-                        costMicroUsd: studentNestedCostMicro || undefined,
-                    }
-                    : undefined,
-                costBreakdown: {
-                    llmRootMicroUsd: studentLlmRootMicro,
-                    nestedSubagentMicroUsd: studentNestedCostMicro,
-                    memoryConsolidationMicroUsd: 0,
-                    compactionMicroUsd: studentCompactionMicro,
+                anyLegUnpriced: anyStudentLegUnpriced,
+                promptTokens: studentPrompt,
+                totalInputTokens: studentTotalInput,
+                cachedTokens: studentCached,
+                outputTokens: studentOutput,
+                cacheWriteTokens: studentCacheWrite,
+                cacheWriteTokensLong: studentCacheWriteLong,
+                llmRootMicroUsd: studentLlmRootMicro,
+                compactionMicroUsd: studentCompactionMicro,
+                nested: {
+                    tokens: studentNestedTokens,
+                    turns: studentNestedTurns,
+                    tasks: studentNestedTasks,
+                    costMicroUsd: studentNestedCostMicro,
                 },
-                humanReview: hrCount > 0 ? { count: hrCount, totalWaitMs: hrWaitMs, gates: hrGates } : undefined,
-            },
+                humanReview: { count: hrCount, totalWaitMs: hrWaitMs, gates: hrGates },
+            }),
             escalations,
             teacherStats: {
                 tokens: teacherTokens,

package/dist/agents/verify.js

@@ -3,6 +3,7 @@ import { CODE_AGENT_PROMPT } from "../prompts/default.js";
 import { releaseSession } from "./session-util.js";
 import { mapNestedSuspend, isDurablePause } from "./suspend-guard.js";
 import { delimitUntrusted, sanitizeUntrustedText } from "../core/untrusted-text.js";
+import { createSafeNotifier } from "../core/safe-notify.js";
 export const VERIFICATION_PROMPT = `You are a verification specialist. Your job is NOT to confirm the implementation works — it is to try to BREAK it.
 
 You have two documented failure patterns. First, verification avoidance: faced with a check, you find reasons not to run it — you read code, narrate what you would test, declare "PASS," and move on. Second, being seduced by the first 80%: a polished result or a passing test suite makes you inclined to pass it, not noticing the edge that crashes, the state that vanishes, the bad input that is unhandled. The first 80% is the easy part. Your entire value is in finding the last 20%.
@@ -58,6 +59,8 @@ export const VerdictSchema = Type.Object({
     }),
     evidence: Type.Optional(Type.String({ description: "Commands run + observed output backing the verdict." })),
 });
+const VERIFY_ON_ROUND_SITE = "verify.onRound";
+const VERIFY_ON_VERIFIER_PAUSE_SITE = "verify.onVerifierPause";
 function fixObjective(findings) {
     const list = findings.length
         ? findings.map((f) => `- ${sanitizeUntrustedText(f)}`).join("\n")
@@ -82,7 +85,8 @@ export async function verifyCompleted(runner, result, specBase, objective, confi
     if (result.status !== "completed") {
         return { ...result, verification: { verdict: "unverified", unverifiedReason: "impl_incomplete", rounds: 0, findings: [] } };
     }
-    const maxRounds = Math.max(1, Math.floor(config.maxRounds ?? 2));
+    const maxRounds = Number.isFinite(config.maxRounds) ? Math.max(1, Math.floor(config.maxRounds)) : 2;
+    const notifier = createSafeNotifier();
     const verifierTools = config.verifierTools ?? (specBase.tools ?? []).filter((t) => t.effect === "read");
     const evidenceMode = config.evidence != null && config.evidence.trim() !== "";
     const verifierPrompt = config.verifierPrompt ?? (evidenceMode ? STATIC_VERIFICATION_PROMPT : VERIFICATION_PROMPT);
@@ -126,6 +130,7 @@ export async function verifyCompleted(runner, result, specBase, objective, confi
         spend += v.cost;
         verifierSpend += v.cost;
         if (v.paused) {
+            const paused = v.paused;
             outcome = {
                 verdict: "unverified",
                 unverifiedReason: "verifier_paused",
@@ -139,18 +144,14 @@ export async function verifyCompleted(runner, result, specBase, objective, confi
                     ...(v.paused.checkpointGate ? { checkpointGate: v.paused.checkpointGate } : {}),
                 },
             };
-            try {
-                config.onVerifierPause?.({
-                    round,
-                    status: v.paused.status,
-                    sessionId: v.paused.sessionId,
-                    ...(v.paused.checkpointToken ? { checkpointToken: v.paused.checkpointToken } : {}),
-                    ...(v.paused.checkpointGate ? { checkpointGate: v.paused.checkpointGate } : {}),
-                });
-            }
-            catch {
-            }
-            config.onRound?.({ round, verdict: outcome.verdict, findings: outcome.findings });
+            notifier.notify(() => config.onVerifierPause?.({
+                round,
+                status: paused.status,
+                sessionId: paused.sessionId,
+                ...(paused.checkpointToken ? { checkpointToken: paused.checkpointToken } : {}),
+                ...(paused.checkpointGate ? { checkpointGate: paused.checkpointGate } : {}),
+            }), VERIFY_ON_VERIFIER_PAUSE_SITE);
+            notifier.notify(() => config.onRound?.({ round, verdict: outcome.verdict, findings: outcome.findings }), VERIFY_ON_ROUND_SITE);
             break;
         }
         const rawVerdict = v.verdict?.verdict ?? "unverified";
@@ -164,7 +165,7 @@ export async function verifyCompleted(runner, result, specBase, objective, confi
             evidence: v.verdict?.evidence,
             verifierCost: verifierSpend,
         };
-        config.onRound?.({ round, verdict: outcome.verdict, findings: outcome.findings });
+        notifier.notify(() => config.onRound?.({ round, verdict: outcome.verdict, findings: outcome.findings }), VERIFY_ON_ROUND_SITE);
         if (outcome.verdict !== "FAIL")
             break;
         if (round >= maxRounds)

package/dist/brain/anthropic.js

@@ -1,7 +1,8 @@
 import {} from "../internal/llm.js";
 import { BrainError } from "./errors.js";
 import { DEGENERATE_MESSAGE, trimDegenerateTail } from "./repetition.js";
-import { createRepetitionPoll, createWalltimeGate } from "./stream-shared.js";
+import { createRepetitionPoll, createWalltimeGate, parseStreamedToolArgs } from "./stream-shared.js";
+import { mintFallbackToolCallId } from "./tool-call-id.js";
 import { WALLTIME_CUTOFF_MESSAGE } from "./walltime.js";
 import { emitBrainTelemetry } from "./status-sink.js";
 import { ANTHROPIC_RESERVED, applyExtraBody, stripAuthHeaders } from "./request-params.js";
@@ -201,23 +202,11 @@ function closeToolUseBlock(acc) {
         return acc.closedTc;
     if (!acc.toolName)
         return undefined;
-    let args;
-    if (acc.toolJson) {
-        try {
-            const parsed = JSON.parse(acc.toolJson);
-            if (typeof parsed !== "object" || parsed === null || Array.isArray(parsed))
-                return undefined;
-            args = parsed;
-        }
-        catch {
-            return undefined;
-        }
-    }
-    else {
-        args = {};
-    }
+    const args = parseStreamedToolArgs(acc.toolJson);
+    if (args === undefined)
+        return undefined;
     if (!acc.toolId)
-        acc.toolId = `toolu_${Math.random().toString(36).slice(2)}`;
+        acc.toolId = mintFallbackToolCallId("toolu_");
     acc.closedTc = { type: "toolCall", id: acc.toolId, name: acc.toolName, arguments: args };
     return acc.closedTc;
 }
@@ -339,6 +328,8 @@ export function createAnthropicBrain(config = {}) {
                 let stopReason = null;
                 let stopDetails;
                 let nextContentIndex = 0;
+                let malformedFrames = 0;
+                let malformedSample;
                 let degenerate = false;
                 let degenTextBlock;
                 const rep = createRepetitionPoll();
@@ -512,6 +503,8 @@ export function createAnthropicBrain(config = {}) {
                             evt = JSON.parse(data);
                         }
                         catch {
+                            malformedFrames++;
+                            malformedSample ??= data.slice(0, 160);
                             return;
                         }
                         handleEvent(evt);
@@ -563,6 +556,12 @@ export function createAnthropicBrain(config = {}) {
                                 text: `\n[note: ${malformed.length} tool call(s) were truncated (stop_reason="${stopReason ?? "?"}") and dropped — re-issue them next turn: ${malformed.join("; ")}]`,
                             });
                         }
+                        if (malformedFrames > 0) {
+                            finalContent.push({
+                                type: "text",
+                                text: `\n[note: ${malformedFrames} provider stream frame(s) were not valid JSON and were dropped — the streamed output may be missing content (first sample: ${JSON.stringify(malformedSample)})]`,
+                            });
+                        }
                         const truncatedEmpty = stopReason === "max_tokens" && noUsableContent && !toolError;
                         const dynamicCut = truncatedEmpty && options?.maxTokensDynamic === true;
                         const staticReasoningCut = truncatedEmpty && !dynamicCut && reasoningSeen && options?.staticReasoningCutDowngrade === true;
@@ -586,7 +585,7 @@ export function createAnthropicBrain(config = {}) {
                                                         ? `[length_empty] response truncated at max_tokens with no answer text (${maxTokensNote}) — the model emitted reasoning although no thinking was requested (a budget-ignoring gateway?) and it consumed the entire output budget; raise max_tokens`
                                                         : `[length_empty] response truncated at max_tokens with no answer text (${maxTokensNote}); raise max_tokens`
                                                 : emptyNoFinish
-                                                    ? `[stream_torn] model stream ended with no content and no stop_reason — the response was lost (a torn stream or an in-band provider error frame)`
+                                                    ? `[stream_torn] model stream ended with no content and no stop_reason — the response was lost (a torn stream or an in-band provider error frame)${malformedFrames > 0 ? ` — ${malformedFrames} unparseable data frame(s) were dropped on this stream (first sample: ${JSON.stringify(malformedSample)})` : ""}`
                                                     : undefined));
                         const errored = walltime.cut ||
                             degenerate ||
@@ -595,9 +594,13 @@ export function createAnthropicBrain(config = {}) {
                             emptyNoFinish ||
                             (toolError !== undefined && noUsableContent) ||
                             (truncatedEmpty && !dynamicCut && !staticReasoningCut);
-                        const errorKind = !walltime.cut && !degenerate && !safetyCut && toolError === undefined && truncatedEmpty && !dynamicCut && !staticReasoningCut
-                            ? "length_empty"
-                            : undefined;
+                        const errorKind = walltime.cut
+                            ? "walltime_cutoff"
+                            : degenerate
+                                ? "degenerate"
+                                : !safetyCut && toolError === undefined && truncatedEmpty && !dynamicCut && !staticReasoningCut
+                                    ? "length_empty"
+                                    : undefined;
                         const finalMessage = {
                             ...partial,
                             content: finalContent.length > 0 ? finalContent : [{ type: "text", text: "" }],