@markusylisiurunen/tau 0.3.93 → 0.3.95

package/dist/core/subagents/agent_supervisor.js

@@ -1,13 +1,10 @@
 import { randomUUID } from "node:crypto";
-import { AgentRuntime, createAgentSpec } from "../agent/agent_runtime.js";
-import { resolveAgentModel } from "../runtime/agent_model.js";
 import { createDefaultCoreDeps } from "../runtime/deps.js";
-import { createAutoCompactionArchiver } from "../session/auto_compaction_archive.js";
-import { ToolCatalog } from "../tools/catalog.js";
-import { appendUsageLogEntry, getUsageCostTotal, getUsageTotals, } from "../usage/logs.js";
-import { extractAssistantText } from "../utils/messages.js";
-import { ClaudeCodeRuntime } from "./claude_code_runtime.js";
+import { appendUsageLogEntry } from "../usage/logs.js";
+import { AgentRequestError } from "./errors.js";
 import { formatActiveSubagentsForCompaction } from "./format.js";
+import { createSubagentRuntime } from "./runtime.js";
+import { AgentTrace } from "./trace.js";
 const MAX_ACTIVE_SUBAGENTS = 8;
 async function raceWithAbort(promise, signal) {
     if (!signal)
@@ -51,7 +48,7 @@ export class AgentSupervisor {
         return [...this.records.values()].filter((record) => record.run.status === "running").length;
     }
     getActiveCompactionContext() {
-        const active = this.listSnapshots().filter((state) => state.availability === "running");
+        const active = this.list().filter((state) => state.status !== "idle");
         if (active.length === 0)
             return undefined;
         const state = formatActiveSubagentsForCompaction(active);
@@ -59,63 +56,25 @@ export class AgentSupervisor {
     }
     spawn(options) {
         if (this.getActiveCount() >= MAX_ACTIVE_SUBAGENTS) {
-            return {
-                ok: false,
-                reason: `Subagent limit reached (max ${MAX_ACTIVE_SUBAGENTS} active). Wait for existing agents to finish.`,
-            };
+            throw new AgentRequestError("operation_failed", `Subagent limit reached (max ${MAX_ACTIVE_SUBAGENTS} active). Wait for existing agents to finish.`);
         }
         const workingDirectory = options.runtimeConfig.workingDirectory.trim();
         if (!workingDirectory) {
-            return { ok: false, reason: "Subagent workingDirectory must not be blank." };
+            throw new AgentRequestError("invalid_arguments", "Subagent workingDirectory must not be blank.");
         }
         const id = randomUUID();
         const runtimeConfig = { ...options.runtimeConfig, workingDirectory };
-        let runtime;
-        let reasoning;
-        if (runtimeConfig.kind === "claude-code") {
-            runtime = new ClaudeCodeRuntime({
-                config: runtimeConfig,
-                backend: options.backend,
-                onEvent: async (event) => await this.recordClaudeCodeEvent(id, event),
-            });
-            reasoning = runtimeConfig.effort;
-        }
-        else {
-            runtime = new AgentRuntime({
-                spec: createAgentSpec({
-                    autoCompactThresholdTokens: null,
-                    ...resolveAgentModel({
-                        id: `${options.personaId ?? "agent"}-subagent`,
-                        label: "subagent",
-                        description: "Background agent",
-                        source: "project",
-                        systemPrompt: runtimeConfig.systemPrompt,
-                        model: runtimeConfig.model,
-                        settings: runtimeConfig.settings,
-                        tools: runtimeConfig.tools,
-                        subagentLaunchModels: [],
-                    }, options.config, {
-                        includeModelNotice: true,
-                        deps: this.deps,
-                    }),
-                    systemPrompt: runtimeConfig.systemPrompt,
-                    tools: ToolCatalog.createSubagentRegistry(runtimeConfig.tools, options.backend, workingDirectory, options.config, options.bashJobs, options.history, options.mcp),
-                }),
-                eventSink: async (event) => await this.recordAgentEvent(id, event),
-                clock: this.deps.clock,
-                archiveAutoCompaction: createAutoCompactionArchiver(options.backend),
-            });
-            reasoning = runtime.spec.attribution.reasoningEffort;
-        }
+        const runtime = createSubagentRuntime(runtimeConfig, {
+            ...options,
+            id,
+            deps: this.deps,
+            onEvent: async (event) => await this.recordRuntimeEvent(id, event),
+        });
         const createdAt = this.deps.clock.now();
         const record = {
             id,
             title: options.title,
-            model: {
-                provider: runtimeConfig.model.provider,
-                id: runtimeConfig.model.id,
-                reasoning,
-            },
+            model: runtime.model,
             workingDirectory,
             createdAt,
             originHistoryEntryId: options.originHistoryEntryId,
@@ -134,157 +93,185 @@ export class AgentSupervisor {
                 cacheRead: 0,
                 cacheWrite: 0,
                 contextWindowUsageTokens: 0,
-                contextWindow: runtimeConfig.kind === "native" ? runtimeConfig.model.contextWindow : 0,
+                contextWindow: runtime.contextWindow,
             },
-            claudeMessageUsage: new Map(),
-            toolPresentations: new Map(),
-            blockedToolCalls: new Map(),
-            completion: Promise.resolve(undefined),
+            completion: Promise.resolve(),
+            trace: new AgentTrace(),
+            pending: [],
         };
         this.records.set(id, record);
+        record.trace.append({
+            kind: "input",
+            mode: "initial",
+            content: [{ type: "text", text: options.prompt }],
+        });
         this.startRun(record, options.prompt, "subagent_spawned");
-        return { ok: true, state: this.toSnapshot(record), capacity: this.getCapacity() };
+        return this.read(record.id);
     }
     sendInput(options) {
-        const record = this.records.get(options.id);
-        if (!record)
-            return { ok: false, reason: `Unknown subagent ID: ${options.id}` };
+        const record = this.requireRecord(options.id);
+        if (record.run.interruptRequested && record.run.status === "running")
+            throw new AgentRequestError("operation_failed", "Subagent is stopping");
+        if (record.run.status !== "running" && this.getActiveCount() >= MAX_ACTIVE_SUBAGENTS)
+            throw new AgentRequestError("operation_failed", "Subagent capacity reached");
+        if (record.pending.length + record.runtime.pendingSteering >= 64)
+            throw new AgentRequestError("operation_failed", "Subagent input queue is full");
+        record.trace.append({
+            kind: "input",
+            mode: options.mode,
+            content: [{ type: "text", text: options.prompt }],
+        });
         if (record.run.status === "running") {
-            return {
-                ok: false,
-                reason: `Subagent ${options.id} is already running. Wait for it to finish before sending input.`,
-            };
+            if (options.mode === "steer")
+                record.runtime.steer(options.prompt);
+            else
+                record.pending.push({ prompt: options.prompt, mode: options.mode });
         }
-        if (this.getActiveCount() >= MAX_ACTIVE_SUBAGENTS) {
-            return {
-                ok: false,
-                reason: `Subagent limit reached (max ${MAX_ACTIVE_SUBAGENTS} active). Wait for existing agents to finish.`,
-            };
+        else {
+            this.beginFollowup(record, options.prompt);
         }
+        return this.read(record.id);
+    }
+    beginFollowup(record, prompt) {
+        record.lastAssistantSeq = undefined;
         record.run = {
             revision: record.run.revision + 1,
             status: "running",
             startedAt: this.deps.clock.now(),
             interruptRequested: false,
         };
-        record.toolPresentations.clear();
-        record.blockedToolCalls.clear();
-        this.startRun(record, options.prompt, "subagent_run_started");
-        return { ok: true, state: this.toSnapshot(record), capacity: this.getCapacity() };
+        this.startRun(record, prompt, "subagent_run_started");
+    }
+    read(id) {
+        const record = this.requireRecord(id);
+        const { run } = record;
+        return {
+            id,
+            title: record.title,
+            model: `${record.model.provider}/${record.model.id}:${record.model.reasoning}`,
+            workingDirectory: record.workingDirectory,
+            ...(run.status === "running"
+                ? { status: run.interruptRequested ? "stopping" : "working" }
+                : run.status === "failed"
+                    ? {
+                        status: "idle",
+                        reason: "failed",
+                        error: { code: run.failure.kind, message: run.failure.message },
+                    }
+                    : { status: "idle", reason: run.status === "succeeded" ? "completed" : "interrupted" }),
+            pending: {
+                queue: record.pending.filter((input) => input.mode === "queue").length,
+                steer: record.runtime.pendingSteering +
+                    record.pending.filter((input) => input.mode === "steer").length,
+            },
+            usage: {
+                input: record.usage.input,
+                output: record.usage.output,
+                cacheRead: record.usage.cacheRead,
+                cacheWrite: record.usage.cacheWrite,
+                contextTokens: record.usage.contextWindowUsageTokens,
+                contextWindow: record.usage.contextWindow,
+                cost: record.costTotal,
+            },
+        };
+    }
+    getTrace(id) {
+        const record = this.requireRecord(id);
+        return record.trace;
     }
     async waitForAgents(ids, signal) {
         const missing = ids.filter((id) => !this.records.has(id));
         if (missing.length > 0)
-            throw new Error(`Unknown subagent ID(s): ${missing.join(", ")}`);
+            throw new AgentRequestError("unknown_agent", `Unknown subagent ID(s): ${missing.join(", ")}`);
         await raceWithAbort(Promise.race(ids.map((id) => this.waitForRecord(id))), signal);
-        return ids.map((id) => {
-            const record = this.records.get(id);
-            if (!record)
-                throw new Error(`Unknown subagent ID: ${id}`);
-            return this.toSnapshot(record);
-        });
+        return ids.map((id) => this.read(id));
     }
-    async interrupt(id, signal) {
-        const record = this.records.get(id);
-        if (!record)
-            return undefined;
+    async interrupt(id, signal, wait = true) {
+        const record = this.requireRecord(id);
         if (record.run.status === "running") {
+            if (record.pending.length || record.runtime.pendingSteering)
+                record.trace.append({
+                    kind: "notice",
+                    code: "input.discarded",
+                    content: [{ type: "text", text: "Pending inputs discarded by interruption." }],
+                });
+            record.pending = [];
             record.run.interruptRequested = true;
+            record.trace.append({ kind: "status", status: "stopping" });
             record.runtime.interrupt();
             await this.emit({ type: "subagent_interrupt_requested", state: this.toSnapshot(record) });
         }
-        return this.toSnapshot(await raceWithAbort(this.waitForRecord(id), signal));
+        if (wait)
+            await raceWithAbort(this.waitForRecord(id), signal);
+        return this.read(id);
+    }
+    getLastResponse(id) {
+        const record = this.requireRecord(id);
+        const seq = record.lastAssistantSeq;
+        if (record.run.status === "running" || seq === undefined)
+            return undefined;
+        try {
+            return { seq, ...record.trace.readText(seq, Number.POSITIVE_INFINITY) };
+        }
+        catch (error) {
+            if (!(error instanceof AgentRequestError) || error.code !== "expired_content")
+                throw error;
+            return { seq, text: "", complete: false };
+        }
     }
     getSnapshot(id) {
         const record = this.records.get(id);
         return record ? this.toSnapshot(record) : undefined;
     }
-    listSnapshots() {
-        return [...this.records.values()].map((record) => this.toSnapshot(record));
-    }
     startRun(record, prompt, eventType) {
-        record.completion = this.emit({ type: eventType, state: this.toSnapshot(record) })
-            .then(async () => {
-            if (record.run.interruptRequested) {
-                this.finishInterrupted(record);
-            }
-            else if (record.runtime instanceof ClaudeCodeRuntime) {
-                const completion = await record.runtime.submit(prompt);
-                if (record.run.interruptRequested || completion.status === "interrupted") {
+        record.trace.append({ kind: "status", status: "working" });
+        record.completion = (async () => {
+            try {
+                await this.emit({ type: eventType, state: this.toSnapshot(record) });
+                if (record.run.interruptRequested) {
                     this.finishInterrupted(record);
                 }
-                else if (completion.result.failed) {
-                    this.finishFailed(record, {
-                        kind: "provider-error",
-                        message: completion.result.response,
-                        stopReason: "error",
-                    });
-                }
                 else {
-                    this.finishSucceeded(record, completion.result.response);
+                    const { outcome, unconsumed } = await record.runtime.submit(prompt);
+                    if (record.run.interruptRequested || outcome.status === "interrupted") {
+                        this.finishInterrupted(record);
+                    }
+                    else {
+                        record.pending.push(...unconsumed.map((prompt) => ({ prompt, mode: "steer" })));
+                        if (outcome.status === "failed")
+                            this.finishFailed(record, outcome.failure);
+                        else
+                            this.finishSucceeded(record, outcome.response);
+                    }
                 }
             }
-            else {
-                this.applyTurnResult(record, await record.runtime.submit(prompt));
-            }
-            return record;
-        })
-            .catch((error) => {
-            if (record.run.interruptRequested) {
-                this.finishInterrupted(record);
+            catch (error) {
+                if (record.run.interruptRequested)
+                    this.finishInterrupted(record);
+                else
+                    this.finishFailed(record, {
+                        kind: "runtime-error",
+                        message: error instanceof Error ? error.message : String(error),
+                    });
             }
-            else {
-                this.finishFailed(record, {
-                    kind: "runtime-error",
-                    message: error instanceof Error ? error.message : String(error),
+            if (this.records.get(record.id) === record) {
+                const next = record.pending.shift();
+                if (next && !record.run.interruptRequested) {
+                    this.beginFollowup(record, next.prompt);
+                    return await record.completion;
+                }
+                record.trace.append({
+                    kind: "status",
+                    status: "idle",
+                    reason: record.run.status === "succeeded"
+                        ? "completed"
+                        : record.run.status === "interrupted"
+                            ? "interrupted"
+                            : "failed",
                 });
+                await this.emit({ type: "subagent_finished", state: this.toSnapshot(record) });
             }
-            return record;
-        })
-            .then(async (resolved) => {
-            if (this.records.get(resolved.id) === resolved) {
-                await this.emit({ type: "subagent_finished", state: this.toSnapshot(resolved) });
-            }
-            return resolved;
-        });
-    }
-    applyTurnResult(record, result) {
-        if (record.run.interruptRequested ||
-            result.aborted ||
-            result.finalMessage?.stopReason === "aborted") {
-            this.finishInterrupted(record);
-            return;
-        }
-        if (result.blocked) {
-            this.finishFailed(record, {
-                kind: result.blocked.reason,
-                message: result.blocked.message,
-            });
-            return;
-        }
-        if (result.limitReached) {
-            this.finishFailed(record, {
-                kind: result.limitReached.reason,
-                message: result.limitReached.message,
-            });
-            return;
-        }
-        if (!result.finalMessage) {
-            this.finishFailed(record, {
-                kind: "runtime-error",
-                message: "Subagent run completed without a final assistant message.",
-            });
-            return;
-        }
-        if (result.finalMessage.stopReason === "error") {
-            this.finishFailed(record, {
-                kind: "provider-error",
-                message: result.finalMessage.errorMessage ?? "model returned an unspecified error.",
-                stopReason: result.finalMessage.stopReason,
-            });
-            return;
-        }
-        this.finishSucceeded(record, extractAssistantText(result.finalMessage).trim());
+        })();
     }
     finishSucceeded(record, response) {
         record.run = {
@@ -307,6 +294,11 @@ export class AgentSupervisor {
         };
     }
     finishFailed(record, failure) {
+        record.trace.append({
+            kind: "notice",
+            code: failure.kind,
+            content: [{ type: "text", text: failure.message }],
+        });
         record.run = {
             revision: record.run.revision,
             status: "failed",
@@ -316,203 +308,57 @@ export class AgentSupervisor {
             failure,
         };
     }
-    async recordClaudeCodeEvent(id, event) {
+    async recordRuntimeEvent(id, event) {
         const record = this.records.get(id);
         if (!record)
             return;
-        if (event.type === "usage" || event.type === "result") {
-            let reported = event.usage;
-            if (event.type === "usage") {
-                const previous = record.claudeMessageUsage.get(event.messageId);
-                // Final result totals already cover these messages, including resumed replays.
-                if (previous === null)
-                    return;
-                const current = {
-                    input: Math.max(previous?.input ?? 0, event.usage.input),
-                    output: Math.max(previous?.output ?? 0, event.usage.output),
-                    cacheRead: Math.max(previous?.cacheRead ?? 0, event.usage.cacheRead),
-                    cacheWrite: Math.max(previous?.cacheWrite ?? 0, event.usage.cacheWrite),
-                };
-                record.claudeMessageUsage.set(event.messageId, current);
-                reported = {
-                    input: record.usage.input + current.input - (previous?.input ?? 0),
-                    output: record.usage.output + current.output - (previous?.output ?? 0),
-                    cacheRead: record.usage.cacheRead + current.cacheRead - (previous?.cacheRead ?? 0),
-                    cacheWrite: record.usage.cacheWrite + current.cacheWrite - (previous?.cacheWrite ?? 0),
-                };
-                if (getUsageTotals(current).total === getUsageTotals(previous).total)
-                    return;
-                record.usage.contextWindowUsageTokens = getUsageTotals(current).total;
-            }
-            else {
-                for (const messageId of record.claudeMessageUsage.keys()) {
-                    record.claudeMessageUsage.set(messageId, null);
-                }
-                record.usage.contextWindowUsageTokens = event.usage.contextWindowUsageTokens;
-                record.usage.contextWindow = event.usage.contextWindow;
-            }
-            // SDK totals include the history restored on resume. Record only the delta,
-            // and replace (rather than sum) the current context-window measurement.
-            const usage = getUsageTotals({
-                input: Math.max(0, reported.input - record.usage.input),
-                output: Math.max(0, reported.output - record.usage.output),
-                cacheRead: Math.max(0, reported.cacheRead - record.usage.cacheRead),
-                cacheWrite: Math.max(0, reported.cacheWrite - record.usage.cacheWrite),
-            });
-            const cost = event.type === "result" ? Math.max(0, event.cost - record.costTotal) : 0;
-            // Startup failures can report zero totals even for a resumed session.
-            record.costTotal += cost;
-            record.usage = {
-                ...record.usage,
-                input: record.usage.input + usage.input,
-                output: record.usage.output + usage.output,
-                cacheRead: record.usage.cacheRead + usage.cacheRead,
-                cacheWrite: record.usage.cacheWrite + usage.cacheWrite,
-            };
-            (this.options.recordUsage ?? appendUsageLogEntry)({
-                timestamp: this.deps.clock.now(),
-                sessionId: record.id,
-                personaId: record.personaId,
-                provider: record.model.provider,
-                model: record.model.id,
-                api: "claude-agent-sdk",
-                reasoningEffort: record.model.reasoning,
-                usage,
-                cost: { total: cost },
-                agent: { type: "subagent" },
-            });
-            await this.emit({ type: "subagent_updated", state: this.toSnapshot(record) });
+        if (event.type === "trace") {
+            const seq = record.trace.append(event.event);
+            if (event.event.kind === "assistant")
+                record.lastAssistantSeq = seq;
         }
-        else if (event.type === "assistant") {
+        else if (event.type === "activity") {
             await this.emit({
                 type: "subagent_activity",
                 state: this.toSnapshot(record),
-                activity: { type: "assistant", text: event.text },
+                activity: event.activity,
             });
         }
-        else if (event.type === "tool") {
-            await this.emit({
-                type: "subagent_activity",
-                state: this.toSnapshot(record),
-                activity: {
-                    type: "tool",
-                    toolName: event.name,
-                    outcome: event.failed ? "failed" : "succeeded",
-                    presentation: {
-                        action: event.failed ? "failed" : "completed",
-                        subject: event.subject,
-                        subjectWrap: "word",
-                        details: [],
-                        metadata: [],
-                    },
-                },
-            });
-        }
-    }
-    async recordAgentEvent(id, event) {
-        const record = this.records.get(id);
-        if (!record)
-            return;
-        if (event.type === "assistant_final") {
-            const usage = getUsageTotals(event.message.usage);
-            const cost = getUsageCostTotal(event.message.usage);
-            record.costTotal += cost;
+        else {
+            const { usage, cost } = event.entry;
+            record.costTotal += cost.total;
             record.usage = {
                 input: record.usage.input + usage.input,
                 output: record.usage.output + usage.output,
                 cacheRead: record.usage.cacheRead + usage.cacheRead,
                 cacheWrite: record.usage.cacheWrite + usage.cacheWrite,
-                contextWindowUsageTokens: usage.input + usage.output + usage.cacheRead + usage.cacheWrite,
-                contextWindow: record.usage.contextWindow,
+                contextWindowUsageTokens: event.contextTokens,
+                contextWindow: event.contextWindow,
             };
             (this.options.recordUsage ?? appendUsageLogEntry)({
-                timestamp: event.message.timestamp,
+                ...event.entry,
                 sessionId: record.id,
-                personaId: record.personaId ?? event.personaId,
-                provider: event.message.provider,
-                model: event.message.model,
-                api: event.message.api,
-                reasoningEffort: event.reasoningEffort,
-                usage,
-                cost: { total: cost },
+                personaId: record.personaId ?? event.entry.personaId,
                 agent: { type: "subagent" },
             });
             await this.emit({ type: "subagent_updated", state: this.toSnapshot(record) });
-            const text = extractAssistantText(event.message).trim();
-            if (text) {
-                await this.emit({
-                    type: "subagent_activity",
-                    state: this.toSnapshot(record),
-                    activity: { type: "assistant", text },
-                });
-            }
-            return;
-        }
-        if (event.type === "tool_activity") {
-            record.toolPresentations.set(event.activity.toolCallId, event.activity.presentation);
-            const blockedToolName = record.blockedToolCalls.get(event.activity.toolCallId);
-            if (blockedToolName) {
-                record.blockedToolCalls.delete(event.activity.toolCallId);
-                await this.emitSettledToolActivity(record, blockedToolName, "blocked", event.activity.presentation);
-                record.toolPresentations.delete(event.activity.toolCallId);
-            }
-            return;
-        }
-        if (event.type === "tool_run_blocked") {
-            const presentation = record.toolPresentations.get(event.toolCallId);
-            if (!presentation) {
-                record.blockedToolCalls.set(event.toolCallId, event.toolName);
-                return;
-            }
-            record.toolPresentations.delete(event.toolCallId);
-            await this.emitSettledToolActivity(record, event.toolName, "blocked", presentation);
-            return;
-        }
-        if (event.type === "tool_run_finished") {
-            const presentation = record.toolPresentations.get(event.toolCallId);
-            if (!presentation) {
-                throw new Error(`missing presentation for settled tool call '${event.toolCallId}'`);
-            }
-            record.toolPresentations.delete(event.toolCallId);
-            await this.emitSettledToolActivity(record, event.toolName, event.outcome, presentation);
-            return;
-        }
-        if (event.type === "feedback") {
-            await this.emit({
-                type: "subagent_activity",
-                state: this.toSnapshot(record),
-                activity: {
-                    type: "notice",
-                    severity: event.tone === "error" ? "error" : "info",
-                    title: event.title,
-                    ...(event.presentation === "transcript" && event.content
-                        ? { content: event.content }
-                        : {}),
-                },
-            });
         }
     }
-    async emitSettledToolActivity(record, toolName, outcome, presentation) {
-        const { actionByStatus, ...terminalPresentation } = presentation;
-        await this.emit({
-            type: "subagent_activity",
-            state: this.toSnapshot(record),
-            activity: {
-                type: "tool",
-                toolName,
-                outcome,
-                presentation: {
-                    action: actionByStatus[outcome],
-                    ...structuredClone(terminalPresentation),
-                },
-            },
-        });
-    }
-    async waitForRecord(id) {
+    requireRecord(id) {
         const record = this.records.get(id);
         if (!record)
-            throw new Error(`Unknown subagent ID: ${id}`);
-        return await record.completion;
+            throw new AgentRequestError("unknown_agent", `Unknown subagent ID: ${id}`);
+        return record;
+    }
+    list() {
+        return [...this.records.keys()].map((id) => this.read(id));
+    }
+    async waitForRecord(id) {
+        const record = this.requireRecord(id);
+        do {
+            await record.completion;
+        } while (record.run.status === "running");
+        return record;
     }
     toSnapshot(record) {
         return {