@herbertgao/pi-subagents 0.15.3 → 0.15.4

package/src/agent-manager.ts

@@ -144,6 +144,38 @@ interface SpawnOptions {
   rootSessionId?: string
 }
 
+interface ResumeOptions {
+  /**
+   * Run the resumed turn detached in the background: return immediately with
+   * the record still "running" (or "queued" at the concurrency limit) and
+   * notify on completion via onComplete, exactly like a background spawn.
+   * Default (false/undefined) runs the resume inline and returns the settled
+   * record — the historical behavior.
+   */
+  isBackground?: boolean
+  /** Called on tool start/end with activity info (for streaming progress to UI). */
+  onToolActivity?: (activity: ToolActivity) => void
+  /** Called at the end of each resumed agentic turn with the cumulative count. */
+  onTurnEnd?: (turnCount: number) => void
+  /** Called once per assistant message_end with that message's usage delta. */
+  onAssistantUsage?: (usage: {
+    input: number
+    output: number
+    cacheWrite: number
+  }) => void
+  /** Called when the session successfully compacts. */
+  onCompaction?: (info: CompactionInfo) => void
+  /**
+   * Background resume only: called synchronously when the run actually starts —
+   * immediately, or later from drainQueue. Callers wire per-run side effects
+   * (output-file streaming) here rather than at the call site, so a resume that
+   * is stopped while still queued never leaves a subscription behind: `abort()`
+   * drops a queued record without reaching `settle()`, which is what would have
+   * torn that subscription down.
+   */
+  onStarted?: () => void
+}
+
 export class AgentManager {
   private agents = new Map<string, AgentRecord>()
   private cleanupInterval: ReturnType<typeof setInterval>
@@ -156,7 +188,7 @@ export class AgentManager {
   private worktreeRepos = new Set<string>()
 
   /** Queue of background agents waiting to start. */
-  private queue: { id: string; args: SpawnArgs }[] = []
+  private queue: { id: string; start: () => void }[] = []
   /** Number of currently running background agents. */
   private runningBackground = 0
 
@@ -236,7 +268,7 @@ export class AgentManager {
       this.runningBackground >= this.maxConcurrent
     ) {
       // Queue it — will be started when a running agent completes
-      this.queue.push({ id, args })
+      this.queue.push({ id, start: () => this.startAgent(id, record, args) })
       return id
     }
 
@@ -508,7 +540,7 @@ export class AgentManager {
       const record = this.agents.get(next.id)
       if (record?.status !== "queued") continue
       try {
-        this.startAgent(next.id, record, next.args)
+        next.start()
       } catch (err) {
         // Late failure (e.g. strict worktree-isolation) — surface on the record
         // so the user/agent can see it via /agents, then keep draining.
@@ -570,10 +602,51 @@ export class AgentManager {
     id: string,
     prompt: string,
     signal?: AbortSignal,
+    options?: ResumeOptions,
   ): Promise<AgentRecord | undefined> {
     const record = this.agents.get(id)
     if (!record?.session) return undefined
 
+    // Background resume: settle asynchronously and notify on completion exactly
+    // like a background spawn, returning immediately with the record still
+    // "running" — or "queued" when at the concurrency limit. Previously
+    // run_in_background was ignored on resume (the Agent tool's resume branch
+    // returned before its background branch, and resume() only ever awaited
+    // inline), so a resumed agent always blocked the caller until it finished.
+    if (options?.isBackground) {
+      // Never re-enter a run that is still in flight. Detaching means the caller
+      // gets control back while the record stays "running", so nothing stops the
+      // model from resuming the same agent again. Starting a second run would
+      // overwrite record.abortController — orphaning the live run beyond the
+      // reach of `/agents` stop and abortAll() — double-count the pool slot, and
+      // then reject from session.prompt() with "Agent is already processing",
+      // whose settle path would abort the LIVE run's children and report a
+      // failure for a run that is still going. Refuse instead, leaving the
+      // record untouched; the caller decides whether to wait or steer.
+      if (record.status === "running" || record.status === "queued")
+        return undefined
+
+      record.isBackground = true
+      record.resultConsumed = false
+      record.result = undefined
+      record.error = undefined
+      record.completedAt = undefined
+      record.status = "queued"
+
+      const start = () => this.startResume(id, record, prompt, signal, options)
+      if (
+        occupiesPoolSlot(record) &&
+        this.runningBackground >= this.maxConcurrent
+      ) {
+        // At the concurrency limit — queue it, drains when a slot frees.
+        this.queue.push({ id, start })
+      } else {
+        start()
+      }
+      return record
+    }
+
+    // Foreground resume: run inline and return the settled record.
     record.status = "running"
     record.startedAt = Date.now()
     record.completedAt = undefined
@@ -584,13 +657,17 @@ export class AgentManager {
       const { text, failure } = await resumeAgent(record.session, prompt, {
         onToolActivity: (activity) => {
           if (activity.type === "end") record.toolUses++
+          options?.onToolActivity?.(activity)
         },
+        onTurnEnd: options?.onTurnEnd,
         onAssistantUsage: (usage) => {
           addUsage(record.lifetimeUsage, usage)
+          options?.onAssistantUsage?.(usage)
         },
         onCompaction: (info) => {
           record.compactionCount++
           this.onCompact?.(record, info)
+          options?.onCompaction?.(info)
         },
         signal,
       })
@@ -613,6 +690,117 @@ export class AgentManager {
     return record
   }
 
+  /**
+   * Start a background resume run: detached, settling and notifying like
+   * startAgent's background path. Invoked immediately, or from drainQueue when
+   * a concurrency slot frees. The session already exists (resume reuses it), so
+   * there is no onSessionCreated to hang per-run wiring off — callers use
+   * `options.onStarted`, which fires on both the immediate and the drained path.
+   */
+  private startResume(
+    id: string,
+    record: AgentRecord,
+    prompt: string,
+    parentSignal: AbortSignal | undefined,
+    options: ResumeOptions,
+  ) {
+    if (!record.session) return
+
+    record.status = "running"
+    record.startedAt = Date.now()
+    if (occupiesPoolSlot(record)) this.runningBackground++
+    this.onStart?.(record)
+
+    // Fresh abort controller so /agents stop and steering target THIS run rather
+    // than the previous one's settled controller.
+    const abortController = new AbortController()
+    record.abortController = abortController
+    // Optional, and NOT what the Agent tool passes for a detached resume: a
+    // parent signal aborts on the parent's own interrupt (user Esc), which is
+    // right for a foreground run whose result the caller is awaiting, and wrong
+    // for a detached one — background spawns omit it for exactly this reason.
+    let detachParentSignal: (() => void) | undefined
+    if (parentSignal) {
+      const onParentAbort = () => this.abort(id)
+      parentSignal.addEventListener("abort", onParentAbort, { once: true })
+      detachParentSignal = () =>
+        parentSignal.removeEventListener("abort", onParentAbort)
+    }
+
+    // Per-run side effects (output streaming) — see ResumeOptions.onStarted.
+    // After the record is in its running shape, before the run is kicked off.
+    try {
+      options.onStarted?.()
+    } catch {
+      /* ignore caller wiring errors */
+    }
+
+    const settle = () => {
+      detachParentSignal?.()
+      detachParentSignal = undefined
+      // Final flush of streaming output file
+      if (record.outputCleanup) {
+        try {
+          record.outputCleanup()
+        } catch {
+          /* ignore */
+        }
+        record.outputCleanup = undefined
+      }
+      // Children spawned during the resumed turn must not outlive it.
+      this.abortOwnedChildren(id)
+      if (occupiesPoolSlot(record)) this.runningBackground--
+      try {
+        this.onComplete?.(record)
+      } catch {
+        /* ignore completion side-effect errors */
+      }
+      this.drainQueue()
+    }
+
+    const promise = resumeAgent(record.session, prompt, {
+      onToolActivity: (activity) => {
+        if (activity.type === "end") record.toolUses++
+        options.onToolActivity?.(activity)
+      },
+      onTurnEnd: options.onTurnEnd,
+      onAssistantUsage: (usage) => {
+        addUsage(record.lifetimeUsage, usage)
+        options.onAssistantUsage?.(usage)
+      },
+      onCompaction: (info) => {
+        record.compactionCount++
+        this.onCompact?.(record, info)
+        options.onCompaction?.(info)
+      },
+      signal: abortController.signal,
+    })
+      .then(({ text, failure }) => {
+        // Don't overwrite status if externally stopped via abort().
+        if (record.status !== "stopped") {
+          // Same contract as the spawn path (#144): a failed final turn is an
+          // error, not a completion — but the resumed text stays available.
+          record.status = failure ? "error" : "completed"
+          if (failure) record.error = failure
+        }
+        record.result = text
+        record.completedAt ??= Date.now()
+        settle()
+        return text
+      })
+      .catch((err) => {
+        if (record.status !== "stopped") {
+          record.status = "error"
+          record.error = err instanceof Error ? err.message : String(err)
+        }
+        record.completedAt ??= Date.now()
+        settle()
+        return ""
+      })
+
+    record.promise = promise
+  }
+
   /**
    * Send a steering message to an agent from the UI (mirrors the steer_subagent
    * tool). A live session delivers it now — it interrupts the agent after its

package/src/agent-runner.ts

@@ -365,7 +365,7 @@ export function setGraceTurns(n: number): void {
  * Try to find the right model for an agent type.
  * Priority: explicit option > config.model > parent model.
  */
-function resolveDefaultModel(
+export function resolveDefaultModel(
   parentModel: Model<any> | undefined,
   registry: {
     find(provider: string, modelId: string): Model<any> | undefined
@@ -946,6 +946,9 @@ export async function runAgent(
     ? SessionManager.create(
         effectiveCwd,
         configuredSessionDir ?? defaultSessionDir,
+        {
+          parentSession: ctx.sessionManager.getSessionFile(),
+        },
       )
     : SessionManager.inMemory(effectiveCwd)
 
@@ -1125,6 +1128,8 @@ export async function resumeAgent(
   prompt: string,
   options: {
     onToolActivity?: (activity: ToolActivity) => void
+    /** Called at the end of each resumed agentic turn with the 1-based count. */
+    onTurnEnd?: (turnCount: number) => void
     onAssistantUsage?: (usage: {
       input: number
       output: number
@@ -1143,10 +1148,18 @@ export async function resumeAgent(
   const startLen = session.messages.length
   const collector = collectResponseText(session)
   const cleanupAbort = forwardAbortSignal(session, options.signal)
+  let turnCount = 0
 
   const unsubEvents =
-    options.onToolActivity || options.onAssistantUsage || options.onCompaction
+    options.onToolActivity ||
+    options.onTurnEnd ||
+    options.onAssistantUsage ||
+    options.onCompaction
       ? session.subscribe((event: AgentSessionEvent) => {
+          if (event.type === "turn_end") {
+            turnCount++
+            options.onTurnEnd?.(turnCount)
+          }
           if (event.type === "tool_execution_start")
             options.onToolActivity?.({
               type: "start",