@gotgenes/pi-subagents 6.0.0 → 6.0.1

package/CHANGELOG.md

@@ -5,6 +5,16 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [6.0.1](https://github.com/gotgenes/pi-packages/compare/pi-subagents-v6.0.0...pi-subagents-v6.0.1) (2026-05-20)
+
+
+### Documentation
+
+* full refresh of architecture.md — reflect completed decomposition ([a8187ce](https://github.com/gotgenes/pi-packages/commit/a8187ce839b465529191b53bbb19fd39d5c71e69))
+* **pi-subagents:** set new architectural target for AgentManager decomposition ([83288da](https://github.com/gotgenes/pi-packages/commit/83288daf3dc6dce7bcad81799a4f6e9f300553c0))
+* plan consolidate test AgentRecord factory ([#102](https://github.com/gotgenes/pi-packages/issues/102)) ([22a3213](https://github.com/gotgenes/pi-packages/commit/22a3213bb55d15871d3582e89d1336fcd9e8fa36))
+* **retro:** add retro notes for issue [#61](https://github.com/gotgenes/pi-packages/issues/61) ([7053be7](https://github.com/gotgenes/pi-packages/commit/7053be7bec86a0c160cb602528d1c59e370ba33d))
+
 ## [6.0.0](https://github.com/gotgenes/pi-packages/compare/pi-subagents-v5.8.2...pi-subagents-v6.0.0) (2026-05-20)
 
 

package/docs/architecture/architecture.md

@@ -1,59 +1,79 @@
 # Architecture
 
-This document describes the planned decomposition of the pi-subagents fork into a focused, composable core with a stable API boundary that other extensions can build on.
+This document describes the architecture of the pi-subagents fork: a focused, composable core with a stable API boundary that other extensions can build on.
 
 ## Design principles
 
 1. **Narrow core** — the extension owns agent spawning, execution, and result retrieval.
    Everything else is a consumer.
 2. **Composable by default** — other extensions can spawn agents, observe their lifecycle, and display their state without importing this package directly.
-3. **Typed API boundary** — this package exports a `SubagentsAPI` interface and `Symbol.for()` accessors (`publishSubagentsAPI` / `getSubagentsAPI`).
+3. **Typed API boundary** — this package exports a `SubagentsService` interface and `Symbol.for()` accessors (`publishSubagentsService` / `getSubagentsService`).
    Consumers declare this package as an optional peer dependency and use dynamic import for compile-time types.
-   The runtime bridge is `Symbol.for()` on `globalThis` — no separate API package.
+   The runtime bridge is `Symbol.for("@gotgenes/pi-subagents:service")` on `globalThis` — no separate API package.
 4. **No scheduling** — in-process scheduling is removed from the core.
    Scheduling is a separate concern that any extension can implement by calling `spawn()` on the published API.
 5. **UI extraction is deferred** — the widget, conversation viewer, and `/agents` command menu stay in the core for now.
    They are the first candidate for extraction once the API boundary is proven stable.
+6. **Snapshot, don't capture** — mutable parent state (ctx, session, model) is read once at spawn time and frozen into a plain data object.
+   No live references survive past the spawn call.
+7. **Subscribe, don't thread** — observation of agent progress uses event subscription on the session, not callback parameters threaded through multiple layers.
 
 ## Current state
 
-The extension is a 6,300 LOC monolith organized into well-factored internal modules but with no public API contract.
-The subsystems are:
+The extension is ~6,100 LOC across 35 focused modules with a typed `SubagentsService` API boundary.
+The `index.ts` entry point is ~270 lines; the rest is decomposed into domain modules.
 
 ```text
-index.ts (1,894 LOC) — entry point, tool registration, event wiring
-agent-manager.ts      — lifecycle, concurrency, queue
-agent-runner.ts       — session creation, turn loop, tool filtering
-agent-types.ts        — type registry (defaults + custom .md files)
-types.ts              — shared type definitions
-
-prompts.ts            — system prompt assembly
-context.ts            — parent conversation extraction
-memory.ts             — persistent MEMORY.md per agent
-skill-loader.ts       — preload .pi/skills into prompts
-env.ts                — git/platform detection
-
-worktree.ts           — git worktree isolation
-usage.ts              — token usage tracking
-model-resolver.ts     — fuzzy model name resolution
-invocation-config.ts  — merge tool params with agent config
-session-dir.ts        — subagent session directory derivation
-settings.ts           — persistent operational settings
-
-cross-extension-rpc.ts — RPC over pi.events                  ← replacing
-group-join.ts         — batch completion notifications
+index.ts (274 LOC)       — entry point, tool registration, event wiring
+agent-manager.ts (499)   — lifecycle, concurrency, queue
+agent-runner.ts (512)    — session creation, turn loop, tool filtering
+session-config.ts (243)  — pure session-config assembler
+agent-types.ts (138)     — type registry (defaults + custom .md files)
+types.ts (126)           — shared type definitions
+runtime.ts (94)          — SubagentRuntime factory (session-scoped state)
+
+prompts.ts               — system prompt assembly
+context.ts               — parent conversation extraction
+memory.ts                — persistent MEMORY.md per agent
+skill-loader.ts          — preload .pi/skills into prompts
+env.ts                   — git/platform detection
+
+worktree.ts              — git worktree isolation
+usage.ts                 — token usage tracking
+model-resolver.ts        — fuzzy model name resolution
+invocation-config.ts     — merge tool params with agent config
+session-dir.ts           — subagent session directory derivation
+settings.ts              — persistent operational settings
+
+service.ts               — SubagentsService interface + Symbol.for() accessors
+service-adapter.ts       — SubagentsService implementation wrapping AgentManager
+
+tools/agent-tool.ts      — Agent tool definition + execute
+tools/get-result-tool.ts — get_subagent_result tool
+tools/steer-tool.ts      — steer_subagent tool
+tools/helpers.ts         — shared tool utilities
+
+handlers/lifecycle.ts    — session_start, session_before_switch, session_shutdown
+handlers/tool-start.ts   — tool_execution_start handler
+
+notification.ts          — completion nudges, custom message renderer
+renderer.ts              — notification TUI component
 
 ui/agent-widget.ts       — above-editor live status widget
+ui/agent-menu.ts         — /agents slash command menu
 ui/conversation-viewer.ts — scrollable session overlay
+
+default-agents.ts        — embedded default agent configs (general-purpose, Explore, Plan)
+custom-agents.ts         — user-defined agent .md file loader
+debug.ts                 — debug logging utility
 ```
 
 ### Coupling today
 
-The widget reads agent state by holding a direct reference to `AgentManager` and polling a shared mutable `Map<string, AgentActivity>` every 80 ms. The conversation viewer subscribes directly to `AgentSession` objects.
-
-Cross-extension consumers use an ad-hoc RPC layer over `pi.events` (`subagents:rpc:spawn`, `subagents:rpc:stop`, `subagents:rpc:ping`) with per-request reply channels and untyped envelopes.
+The widget reads agent state by holding a direct reference to `SubagentRuntime` and polling a shared mutable `Map<string, AgentActivity>` every 80 ms. The conversation viewer subscribes directly to `AgentSession` objects.
 
-There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that exposes `{ waitForAll, hasRunning, spawn, getRecord }`, but it is undocumented and untyped.
+Cross-extension consumers use the typed `SubagentsService` API published via `Symbol.for("@gotgenes/pi-subagents:service")` on `globalThis`.
+The ad-hoc RPC layer and untyped `Symbol.for("pi-subagents:manager")` have been removed.
 
 ## Target state
 
@@ -62,20 +82,20 @@ There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that
   │  @gotgenes/pi-subagents  (this package)                 │
   │                                                        │
   │  Exports:                                              │
-  │    SubagentsAPI interface                              │
-  │    publishSubagentsAPI() / getSubagentsAPI()           │
-  │    SubagentRecord, SubagentStatus, LifetimeUsage types │
-  │    Event channel constants                             │
+  │    SubagentsService interface                           │
+  │    publishSubagentsService() / getSubagentsService()    │
+  │    SubagentRecord, SubagentStatus, LifetimeUsage types  │
+  │    SUBAGENT_EVENTS constants                            │
   │                                                        │
   │  Core:                                                 │
   │    Agent + get_subagent_result + steer_subagent tools  │
   │    AgentManager, agent-runner, agent-types             │
-  │    publishSubagentsAPI(impl)  ← called at init         │
+  │    publishSubagentsService(impl)  ← called at init     │
   │                                                        │
   │  Internal UI (widget, viewer, /agents menu)            │
   │  ← moves to pi-subagents-ui later                     │
   └──────────────────────┬─────────────────────────────────┘
-                         │ Symbol.for("pi:service:subagents")
+                         │ Symbol.for("@gotgenes/pi-subagents:service")
                          │
        ┌─────────────────┼──────────────────┐
        │                 │                  │
@@ -87,7 +107,7 @@ There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that
   │  ext)   │    └──────────────┘    └──────────────┘
   └─────────┘
        │
-       │  getSubagentsAPI()?.spawn(...)
+       │  getSubagentsService()?.spawn(...)
        │  (optional peer dep + dynamic import for types)
        ▼
 ```
@@ -97,10 +117,13 @@ There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that
 - The three tools: `Agent`, `get_subagent_result`, `steer_subagent`.
 - `AgentManager` — spawn, queue, abort, resume, concurrency control.
 - `agent-runner` — session creation, turn loop, tool filtering, extension binding (Patches 2 and 3).
+- `session-config` — pure configuration assembler (extracted from `agent-runner`).
+- `SubagentRuntime` — session-scoped state bag with methods.
 - Agent type registry — default agents, custom `.md` file loading.
 - Prompt assembly, context extraction, memory, skills, environment.
 - Worktree isolation.
 - Token usage tracking.
+- Session directory derivation and persisted `SessionManager` for subagent transcripts.
 - Settings persistence.
 - Internal UI (widget, conversation viewer, `/agents` menu) — these stay until the API boundary is proven, then move to a separate extension.
 
@@ -108,8 +131,8 @@ There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that
 
 - **Scheduling** (`schedule.ts`, `schedule-store.ts`, `ui/schedule-menu.ts`) — 612 LOC removed.
   The `schedule` parameter is removed from the `Agent` tool schema.
-  Any extension that wants scheduling can implement it by calling `getSubagentsAPI()?.spawn(...)` on a timer.
-- **Ad-hoc RPC** (`cross-extension-rpc.ts`) — replaced by the typed `SubagentsAPI` published via `Symbol.for()`.
+  Any extension that wants scheduling can implement it by calling `getSubagentsService()?.spawn(...)` on a timer.
+- **Ad-hoc RPC** (`cross-extension-rpc.ts`) — replaced by the typed `SubagentsService` published via `Symbol.for()`.
   The untyped event-bus RPC channels are removed.
 - **Group join** (`group-join.ts`) — 141 LOC removed.
   The grouped notification batching adds complexity for a marginal UX improvement.
@@ -117,21 +140,22 @@ There is also a `Symbol.for("pi-subagents:manager")` export on `globalThis` that
 - **Output file** (`output-file.ts`) — replaced by `session-dir.ts` + `SessionManager.create()` (#61).
   Subagent transcripts are now written in Pi's official JSONL session format via the SDK's `SessionManager`, nested under the parent session directory.
 
-### Estimated impact
+### Estimated impact (realized)
 
-| Subsystem removed | LOC removed   | LOC removed from index.ts |
-| ----------------- | ------------- | ------------------------- |
-| Scheduling        | 612           | ~200                      |
-| Ad-hoc RPC        | 80            | ~50                       |
-| Group join        | 141           | ~100                      |
-| Output file       | 83 (replaced) | ~50                       |
-| **Total**         | **~916**      | **~400**                  |
+| Subsystem              | Status         | LOC impact                                 |
+| ---------------------- | -------------- | ------------------------------------------ |
+| Scheduling             | Removed (#52)  | −612                                       |
+| Ad-hoc RPC             | Removed (#49)  | −080                                       |
+| Group join             | Removed (#49)  | −141                                       |
+| Output file            | Replaced (#61) | −83 (replaced by 38-line `session-dir.ts`) |
+| index.ts decomposition | Done (#54)     | 1,894 → 274                                |
 
-After removal and `index.ts` decomposition, the core shrinks from ~6,300 to ~5,400 LOC, and `index.ts` shrinks from ~1,894 to ~1,300 LOC.
+The codebase is now ~6,100 LOC across 35 modules.
+The `index.ts` entry point is 274 lines.
 
-## SubagentsAPI
+## SubagentsService (done — #48)
 
-The `SubagentsAPI` interface, accessor functions, and serializable types are exported directly from this package (`@gotgenes/pi-subagents`).
+The `SubagentsService` interface, accessor functions, and serializable types are exported from `@gotgenes/pi-subagents` via the `./service` export map entry.
 No separate API package is needed.
 
 Consumers declare this package as an optional peer dependency:
@@ -139,7 +163,7 @@ Consumers declare this package as an optional peer dependency:
 ```json
 {
   "peerDependencies": {
-    "@gotgenes/pi-subagents": ">=2.0.0"
+    "@gotgenes/pi-subagents": ">=5.0.0"
   },
   "peerDependenciesMeta": {
     "@gotgenes/pi-subagents": { "optional": true }
@@ -150,72 +174,40 @@ Consumers declare this package as an optional peer dependency:
 At runtime, consumers use dynamic import for type-safe access to the accessor functions:
 
 ```typescript
-const { getSubagentsAPI } = await import("@gotgenes/pi-subagents");
-const api = getSubagentsAPI();
-if (api) {
-  api.spawn("Explore", "Check for stale TODOs");
+const { getSubagentsService } = await import("@gotgenes/pi-subagents");
+const svc = getSubagentsService();
+if (svc) {
+  svc.spawn("Explore", "Check for stale TODOs");
 }
 ```
 
 Pi's extension loader creates a fresh `jiti` instance per extension with `moduleCache: false`, so module-scoped singletons don't survive across extensions.
-The accessor functions use `Symbol.for()` on `globalThis`, which is process-global by spec, to bridge this gap.
+The accessor functions use `Symbol.for("@gotgenes/pi-subagents:service")` on `globalThis`, which is process-global by spec, to bridge this gap.
 The dynamic import provides compile-time types; the `Symbol.for()` key is the actual runtime channel.
 
 ### Interface
 
-```typescript
-/** The public API surface published by pi-subagents. */
-export interface SubagentsAPI {
-  /**
-   * Spawn an agent. Returns the agent ID immediately.
-   * The agent runs in the background unless options.foreground is true.
-   */
-  spawn(type: string, prompt: string, options?: SpawnOptions): string;
-
-  /** Get a snapshot of an agent's current state. */
-  getRecord(id: string): SubagentRecord | undefined;
-
-  /** List all tracked agents, most recent first. */
-  listAgents(): SubagentRecord[];
-
-  /** Abort a running or queued agent. Returns false if not found. */
-  abort(id: string): boolean;
-
-  /** Send a steering message to a running agent. */
-  steer(id: string, message: string): Promise<boolean>;
+See `src/service.ts` for the canonical definition.
+Key types:
 
-  /** Wait for all running and queued agents to complete. */
-  waitForAll(): Promise<void>;
-
-  /** Whether any agents are running or queued. */
-  hasRunning(): boolean;
-}
-
-export interface SpawnOptions {
-  description?: string;
-  model?: string;
-  maxTurns?: number;
-  thinkingLevel?: string;
-  isolated?: boolean;
-  inheritContext?: boolean;
-  foreground?: boolean;
-  /** Skip the concurrency queue — start immediately. */
-  bypassQueue?: boolean;
-  isolation?: "worktree";
-}
-```
+- `SubagentsService` — `spawn`, `getRecord`, `listAgents`, `abort`, `steer`, `waitForAll`, `hasRunning`.
+- `SubagentRecord` — serializable agent snapshot (no live session objects).
+- `SpawnOptions` — `description`, `model`, `maxTurns`, `thinkingLevel`, `isolated`, `inheritContext`, `foreground`, `bypassQueue`, `isolation`.
+- `SUBAGENT_EVENTS` — channel constants for `pi.events` subscriptions.
 
 ### Accessor pattern
 
 ```typescript
-const KEY = Symbol.for("pi:service:subagents");
+const SERVICE_KEY = Symbol.for("@gotgenes/pi-subagents:service");
 
-export function publishSubagentsAPI(api: SubagentsAPI): void {
-  (globalThis as any)[KEY] = api;
+export function publishSubagentsService(service: SubagentsService): void {
+  (globalThis as Record<symbol, unknown>)[SERVICE_KEY] = service;
 }
 
-export function getSubagentsAPI(): SubagentsAPI | undefined {
-  return (globalThis as any)[KEY];
+export function getSubagentsService(): SubagentsService | undefined {
+  return (globalThis as Record<symbol, unknown>)[SERVICE_KEY] as
+    | SubagentsService
+    | undefined;
 }
 ```
 
@@ -237,23 +229,19 @@ They are fire-and-forget broadcast events — no request IDs, no reply channels.
 ### Consumer example: scheduling extension
 
 ```typescript
-// package.json:
-// "peerDependencies": { "@gotgenes/pi-subagents": ">=2.0.0" }
-// "peerDependenciesMeta": { "@gotgenes/pi-subagents": { "optional": true } }
-
 export default function (pi) {
   pi.on("session_start", async (event, ctx) => {
-    let getSubagentsAPI;
+    let getSubagentsService;
     try {
-      ({ getSubagentsAPI } = await import("@gotgenes/pi-subagents"));
+      ({ getSubagentsService } = await import("@gotgenes/pi-subagents"));
     } catch {
       return; // pi-subagents not installed
     }
-    const api = getSubagentsAPI();
-    if (!api) return;
+    const svc = getSubagentsService();
+    if (!svc) return;
 
     setInterval(() => {
-      api.spawn("Explore", "Check for stale TODOs", {
+      svc.spawn("Explore", "Check for stale TODOs", {
         bypassQueue: true,
       });
     }, 60 * 60 * 1000);
@@ -267,13 +255,13 @@ export default function (pi) {
 export default function (pi) {
   pi.events.on("subagents:completed", async (data) => {
     const { id } = data as { id: string };
-    let getSubagentsAPI;
+    let getSubagentsService;
     try {
-      ({ getSubagentsAPI } = await import("@gotgenes/pi-subagents"));
+      ({ getSubagentsService } = await import("@gotgenes/pi-subagents"));
     } catch {
       return;
     }
-    const record = getSubagentsAPI()?.getRecord(id);
+    const record = getSubagentsService()?.getRecord(id);
     if (record?.result) {
       fs.appendFileSync("agent-log.jsonl", JSON.stringify(record) + "\n");
     }
@@ -281,32 +269,38 @@ export default function (pi) {
 }
 ```
 
-## index.ts decomposition
+## index.ts decomposition (done — #54, #69, #70)
 
-The 1,894-line `index.ts` is decomposed into focused modules:
+The original 1,894-line `index.ts` has been decomposed into focused modules:
 
 ```text
 src/
-├── index.ts                  ← slimmed entry point: init, tool registration
+├── index.ts (274)            ← slimmed entry point: init, tool registration
+├── runtime.ts (94)           ← SubagentRuntime: session-scoped state + methods
 ├── tools/
-│   ├── agent-tool.ts         ← Agent tool definition + execute
-│   ├── result-tool.ts        ← get_subagent_result tool
-│   └── steer-tool.ts         ← steer_subagent tool
-├── notifications.ts          ← completion nudges, custom renderer
-├── activity-tracker.ts       ← AgentActivity map + callback factory
-├── agents-command.ts         ← /agents slash command menu
-├── api-adapter.ts            ← SubagentsAPI implementation wrapping AgentManager
-└── (existing modules unchanged)
+│   ├── agent-tool.ts (626)   ← Agent tool definition + execute
+│   ├── get-result-tool.ts    ← get_subagent_result tool
+│   ├── steer-tool.ts         ← steer_subagent tool
+│   └── helpers.ts            ← shared tool utilities
+├── handlers/
+│   ├── lifecycle.ts          ← session_start, session_before_switch, session_shutdown
+│   └── tool-start.ts         ← tool_execution_start handler
+├── notification.ts           ← completion nudges, custom renderer
+├── renderer.ts               ← notification TUI component
+├── ui/agent-menu.ts (677)    ← /agents slash command menu
+├── service-adapter.ts        ← SubagentsService implementation wrapping AgentManager
+└── (existing domain modules unchanged)
 ```
 
 Each extracted module receives narrow constructor-injected dependencies rather than closing over module-level state.
+Handlers call methods on narrow runtime interfaces — no raw field writes, no `widget!` reach-throughs.
 
-## Phase plan
+## Phase plan (Phases 1–5 complete)
 
-### Phase 1: Export `SubagentsAPI` from this package
+### Phase 1: Export `SubagentsService` from this package ✓ (done — #48)
 
-Add the `SubagentsAPI` interface, serializable types, and `Symbol.for()` accessor functions as public exports of this package.
-No behavioral changes to the core yet.
+Added the `SubagentsService` interface, serializable types, `Symbol.for()` accessor functions, and `SUBAGENT_EVENTS` constants as public exports.
+Wired `service-adapter.ts` to wrap `AgentManager` and call `publishSubagentsService()` at extension init.
 
 ### Phase 2: Remove scheduling ✓ (done — issue #52)
 
@@ -314,127 +308,225 @@ Deleted `schedule.ts`, `schedule-store.ts`, `ui/schedule-menu.ts`.
 Removed the `schedule` parameter from the `Agent` tool schema.
 Removed scheduler setup and lifecycle hooks from `index.ts`.
 
-### Phase 3: Remove group-join, ad-hoc RPC; replace output-file
+### Phase 3: Remove group-join, ad-hoc RPC; replace output-file ✓ (done — #49, #61)
 
-Delete `group-join.ts`, `cross-extension-rpc.ts`.
-Replace `output-file.ts` with `SessionManager.create()` + `session-dir.ts` (#61).
-Simplify `index.ts` to use direct individual notifications.
-Emit lifecycle events on `pi.events` for external consumers.
+Deleted `group-join.ts`, `cross-extension-rpc.ts` (#49).
+Replaced `output-file.ts` with `SessionManager.create()` + `session-dir.ts` (#61).
+Simplified `index.ts` to use direct individual notifications.
+Lifecycle events emitted on `pi.events` for external consumers.
 
-### Phase 4: Implement and publish `SubagentsAPI`
+### Phase 4: Implement and publish `SubagentsService` ✓ (done — #48)
 
-Wire `api-adapter.ts` to wrap `AgentManager` and call `publishSubagentsAPI()` at extension init.
-Resolve model strings inside the adapter (fixing upstream [tintinweb/pi-subagents#60]).
+Wired `service-adapter.ts` to wrap `AgentManager` and call `publishSubagentsService()` at extension init.
+Model strings are resolved inside the adapter.
 
-### Phase 5: Decompose `index.ts` ✓ (done — issue #54)
+### Phase 5: Decompose `index.ts` ✓ (done — #54, #69, #70, #87)
 
-Extracted tools, notifications, activity tracking, and the `/agents` command into separate modules.
-`src/index.ts` shrank from ~1,619 lines to ~265 lines.
+Extracted tools, notifications, activity tracking, event handlers, and the `/agents` command into separate modules.
+Created `SubagentRuntime` factory to hold session-scoped state.
+`src/index.ts` shrank from ~1,894 lines to ~274 lines.
 
 ### Phase 6 (future): Extract UI to `@gotgenes/pi-subagents-ui`
 
-Move `ui/agent-widget.ts`, `ui/conversation-viewer.ts`, the `/agents` command, notifications, and activity tracking to a separate extension that consumes `SubagentsAPI` + lifecycle events.
+Move `ui/agent-widget.ts`, `ui/conversation-viewer.ts`, the `/agents` command, notifications, and activity tracking to a separate extension that consumes `SubagentsService` + lifecycle events.
 This phase is deferred until the API boundary is proven stable in production.
 
-## Structural refactoring roadmap (post-#54)
+## Structural refactoring roadmap (post-#54) ✓ complete
 
-The Issue #54 decomposition created focused modules but left several structural cleanup opportunities on the table.
-The following issues track the work needed to bring `pi-subagents` to the same level of testability and composability as `pi-permission-system`.
+All structural refactoring phases are complete.
+See `git log` for the full history; issue references are preserved below for traceability.
 
-### Phase 1: Foundation
+| Phase              | Issue              | Summary                                                               |
+| ------------------ | ------------------ | --------------------------------------------------------------------- |
+| Foundation         | #69, #71, #76, #80 | SubagentRuntime, pure assembler, cwd injection, config consolidation  |
+| Core decomposition | #84, #72, #87, #70 | WorktreeManager, AgentManager DI, runtime methods, handler extraction |
+| Interface polish   | #66, #77           | SDK types, projectAgentsDir                                           |
+| Features           | #61                | JSONL session transcripts                                             |
 
-These issues are independent of each other and can land in any order.
-Together they eliminate module-scope mutable state, create a testable functional core, and simplify the agent-types API.
+The remaining open issue is #22 (parent-session resolution), a cross-extension track that does not gate the structural work.
 
-1. **gotgenes/pi-packages#69** ✓ — Create `SubagentRuntime`
-   - Move `defaultMaxTurns`, `graceTurns`, `agentActivity`, `currentCtx`, and widget references out of closure/module scope into a single factory-constructed object.
-   - This unblocks handler extraction (Issue #70) by giving handlers a concrete deps bag instead of closure variables.
+---
 
-2. **gotgenes/pi-packages#71** ✓ — Extract pure agent-session assembler from `agent-runner.ts`
-   - Split `runAgent()` into a pure configuration assembler (~200 lines) and an IO shell (~200 lines).
-   - The assembler becomes independently testable without mocking the Pi SDK.
+## Next target: AgentManager internal decomposition
 
-3. **gotgenes/pi-packages#76** ✓ — Inject `cwd` into `AgentManager`
-   - Replaced the `process.cwd()` call in `dispose()` with a constructor parameter.
+The structural refactoring roadmap decomposed the extension entry point and established clean module boundaries.
+AgentManager itself — the central class — was not touched structurally.
+A design review reveals three tangled responsibilities and two systemic patterns that inflate complexity.
 
-4. **gotgenes/pi-packages#80** ✓ — Consolidate `getConfig` / `getAgentConfig` into a single resolution path
-   - Replaced the two overlapping lookup functions with a single `resolveAgentConfig(type): AgentConfig` that handles the unknown-type fallback internally.
-   - Eliminated the duplicated fallback chain exposed by #71 and simplified test mock setup.
+### Problem statement
 
-### Phase 2: Core decomposition
+AgentManager is a 500-line class that serves as the single mediator between tool callers and the agent runner.
+Every concern passes through it because it owns the `AgentRecord`.
 
-These build on Phase 1 and should land after it.
+Three responsibilities are tangled:
 
-1. **gotgenes/pi-packages#84** ✓ — Extract `GitWorktreeManager` class from `worktree.ts`
-   - Added `WorktreeManager` interface and `GitWorktreeManager` class that captures `cwd` at construction.
-   - Prerequisite for #72 — separated the real-object extraction from the DI refactor.
+1. **Record registry** — create, track, query, clean up `AgentRecord` instances.
+2. **Concurrency control** — queue, running count, drain, `bypassQueue`.
+3. **Execution orchestration** — thread options to the runner, intercept callbacks to update records, wire abort signals, manage worktree lifecycle.
 
-2. **gotgenes/pi-packages#72** ✓ — Dependency-inject `AgentManager`'s collaborators
-   - Defined `AgentRunner` interface (execution boundary) and `ResumeOptions` type in `agent-runner.ts`.
-   - Converted `AgentManager` constructor from 6 positional parameters to an `AgentManagerOptions` bag with injected `AgentRunner` and `WorktreeManager`.
-   - Removed all runtime imports of `agent-runner.ts` and `worktree.ts` from `agent-manager.ts` (only `import type` remains).
-   - Migrated all tests from `vi.mock()` module stubs to `vi.fn()` interface stubs.
+`startAgent()` alone is ~130 lines because it handles all three.
+The `.then()` / `.catch()` blocks mix status updates (job 1), worktree cleanup (job 3), notification callbacks (job 1), and queue draining (job 2).
 
-3. **gotgenes/pi-packages#87** ✓ — Evolve `SubagentRuntime` from data bag to object with methods
-   - Added session-context methods (`setSessionContext`, `clearSessionContext`) and widget delegation methods (`setUICtx`, `onTurnStart`, `markFinished`, `updateWidget`, `ensureTimer`).
-   - Prerequisite for #70 — without runtime methods, extracted handlers would move LoD violations and output-argument smells into handler classes.
+Two systemic patterns compound the problem:
 
-4. **gotgenes/pi-packages#70** ✓ — Extract event handlers into `src/handlers/`
-   - Moved the four inline lambdas (`session_start`, `session_before_switch`, `session_shutdown`, `tool_execution_start`) into `SessionLifecycleHandler` and `ToolStartHandler` classes.
-   - Handlers call methods on narrow runtime interfaces — no raw field writes, no `widget!` reach-throughs.
+### Problem 1: Callback threading
 
-### Phase 3: Interface polish
+`SpawnOptions` carries 6 `on*` callback fields.
+They thread through three layers:
 
-Small cleanups that are safest after the structural changes settle.
+```text
+agent-tool.ts (UI tracking state)
+  → AgentManager.startAgent() wraps each to update the record, then forwards
+    → runner.run() subscribes to session events, calls callbacks
+```
 
-1. **gotgenes/pi-packages#66** — Replace `as any` casts with proper SDK types
-   - Type-only change in the tool/menu factory dep interfaces.
-   - Best done after Issues #69 and #70 when the interfaces are stable.
+The callbacks serve two purposes that are tangled together:
 
-2. **gotgenes/pi-packages#77** — Add `projectAgentsDir` to `AgentMenuDeps`
-   - Remove the inline `process.cwd()` lambda from the menu handler.
+1. **Record statistics** — `onToolActivity` increments `toolUses`, `onAssistantUsage` accumulates `lifetimeUsage`, `onCompaction` increments `compactionCount`, `onSessionCreated` captures the session and output file.
+   This is internal bookkeeping that belongs to the record.
+2. **UI streaming** — the same callbacks update the widget's active-tool display, response text preview, and turn counter.
+   This is presentation that belongs to the UI layer.
 
-### Phase 4: Features and cross-cutting concerns
+The session already emits all of these events via `session.subscribe()`.
+The runner subscribes to session events, translates them into callback invocations, AgentManager wraps each callback to update the record, then forwards to the caller's callback.
+Three layers reimplementing what a single event subscription could provide.
 
-1. **gotgenes/pi-packages#61** ✓ — Port transcript logging to Pi's official JSONL session format
-   - Replaced `output-file.ts` with `SessionManager.create()` + `session-dir.ts`.
-   - Subagent sessions are persisted under `<parent-session-dir>/<parent-session-basename>/tasks/` with `parentSession` header linking.
+### Problem 2: Live `ctx` capture
 
-2. **gotgenes/pi-packages#22** — Parent-session resolution for `nicobailon/pi-subagents` children
-   - Cross-extension issue that spans `pi-permission-system` and `pi-subagents`.
-   - Requires coordination on env-var conventions.
-   - Not blocked by the structural refactor but logically separate from it.
+`ctx: ExtensionContext` is a mutable reference to the parent session.
+It is captured into `SpawnArgs` and held in the concurrency queue:
 
-### Dependency graph
+```typescript
+const args: SpawnArgs = { pi, ctx, type, prompt, options };
+this.queue.push({ id, args });  // ctx held until dequeue
+```
 
-```text
-#69 (SubagentRuntime) ✓ ──► #87 (runtime methods) ✓ ─┬─► #70 (handler extraction) ✓
-                                                   │
-#71 (pure assembler) ✓                              │
-#80 (config lookup) ✓                               │
-#76 (cwd injection) ✓                               │
-#84 (WorktreeManager) ✓                             │
-#72 (AgentManager DI) ✓ ────────────────────────────┘──(optional)──► #70
-
-#66 (type casts) ◄─────(after structural changes settle)
-#77 (projectAgentsDir) ◄─(after #66 or parallel)
-
-#61 (transcript format) ✓
-#22 (parent session) ◄──(cross-extension, independent)
+When the queued agent dequeues, `runAgent()` reads from the live `ctx`:
+
+- `ctx.cwd` — directory that may have changed.
+- `ctx.getSystemPrompt()` — live method call on a potentially stale session.
+- `ctx.model` — model that may have been switched.
+- `ctx.modelRegistry` — registry reference.
+
+If the parent session changes between queue and dequeue (model switch, cwd change, session restart), the agent reads invalid state.
+The same live reference persists in `runtime.currentCtx` for the service-adapter.
+
+Additionally, `inheritContext` calls `ctx.sessionManager.getBranch()` at run time.
+The user's intent is to fork the conversation as it existed when they asked for the agent — not the conversation at some arbitrary later point when a queue slot opens.
+
+### Design: snapshot at spawn time
+
+Replace the live `ctx` capture with a plain data snapshot taken once at spawn time:
+
+```typescript
+interface ParentSnapshot {
+  cwd: string;
+  systemPrompt: string;
+  model: unknown;
+  modelRegistry: { find(...): unknown; getAvailable?(): ... };
+  parentContext?: string;  // pre-built text if inheritContext
+}
 ```
 
-### Recommended order
+This snapshot is:
 
-The recommended sequence is:
+- Captured once in `spawn()` (or by the tool before calling `spawn()`).
+- Stored in `SpawnArgs` instead of `ctx`.
+- Passed to `runner.run()` instead of `ctx: ExtensionContext`.
+- Immutable — no staleness risk, no session-lifetime coupling.
+
+`runAgent()` already reads exactly these 4 values from `ctx` and never touches it again.
+`buildParentContext()` also reads once and produces a string.
+The snapshot formalizes what is already happening, and makes the "read once" guarantee structural.
+
+### Design: session-event observation replaces callback threading
+
+The session emits events via `session.subscribe()`.
+Today, `runner.run()` subscribes and translates events into `RunOptions.on*()` callbacks, AgentManager wraps those to update the record, then forwards to the caller.
+
+The target replaces this three-layer chain with direct subscription:
 
 ```text
-#69 ✓ → #71 ✓ → #80 ✓ → #76 ✓ → #84 ✓ → #72 ✓ → #87 ✓ → #70 ✓ → #66 → #77 → #61 ✓
+                     session.subscribe()
+                            │
+              ┌─────────────┼─────────────┐
+              │                           │
+       Record observer              UI observer
+  (accumulates stats on record)   (updates widget state)
+  managed by AgentManager         managed by agent-tool
+  subscribes in startAgent()      subscribes after spawn
 ```
 
-Phase 1 is complete; Phase 2 is complete.
-Issue #61 (transcript format) is complete.
-The next issue is #66 (replace `as any` casts with proper SDK types).
-Issue #22 is a parallel cross-extension track and does not gate the structural work.
+AgentManager subscribes to the session to update the record (toolUses, lifetimeUsage, compactionCount, outputFile).
+The agent-tool subscribes to the session to stream UI state (active tools, response text, turn count).
+Neither layer wraps or forwards the other's callbacks.
+
+`RunOptions` drops all 6 `on*` fields and becomes pure configuration.
+`SpawnOptions` drops all 6 `on*` fields and becomes identity + dispatch mode.
+The session reference reaches callers via `record.session` (already stored) or via an `onSessionCreated` callback that is the one callback that remains (it delivers the session object, enabling the external subscription).
+
+### Design: record state machine
+
+Status transitions are scattered across 6 locations (`startAgent` `.then()`, `.catch()`, `resume()`, `abort()`, `abortAll()`, `drainQueue()`).
+Each location sets `record.status` plus associated fields (`completedAt`, `result`, `error`) in ad-hoc combinations.
+
+Extract a state machine on `AgentRecord` (or a thin wrapper) that owns all transitions:
+
+```typescript
+record.markRunning(startedAt)
+record.markCompleted(result, completedAt)
+record.markError(error)
+record.markStopped()
+record.resetForResume()
+```
+
+Each method sets exactly the fields that belong to that transition.
+Invalid transitions (e.g., `markCompleted` on an already-stopped record) are no-ops.
+The `if (record.status !== "stopped")` guards in `.then()` and `.catch()` become part of the transition logic rather than scattered conditionals.
+
+### Phased implementation
+
+The three designs are independent and can land in any order.
+The recommended sequence minimizes intermediate churn.
+
+#### Step 1: Record state machine
+
+Extract status-transition methods onto `AgentRecord` (or a `RecordManager` wrapper).
+Purely mechanical — replace scattered field writes with method calls.
+No interface changes for callers.
+
+This is the lowest-risk change and immediately reduces `startAgent()` line count.
+
+#### Step 2: Parent snapshot
+
+Replace `ctx: ExtensionContext` in `SpawnArgs` with a `ParentSnapshot` data object.
+Capture the snapshot in `spawn()` or at the tool call site.
+Update `runner.run()` signature to accept `ParentSnapshot` instead of `ctx`.
+Remove `pi: ExtensionAPI` from `SpawnArgs` (it is only used to pass to `runner.run()`, which only uses it for `detectEnv()` — that can accept a shell-exec function instead).
+
+This change narrows the `AgentRunner` interface and eliminates live-reference capture.
+
+#### Step 3: Session-event observation
+
+Replace the callback-threading pattern with direct session subscriptions.
+AgentManager subscribes to the session after creation to update the record.
+The agent-tool subscribes to the session after spawn to stream UI state.
+`RunOptions` and `SpawnOptions` drop all `on*` callback fields.
+
+This is the largest change but depends on Step 2 (the runner signature is already narrower) and benefits from Step 1 (the record's transition methods encapsulate the stats updates that the subscription drives).
+
+### Expected outcome
+
+| Metric                            | Before | After                    |
+| --------------------------------- | ------ | ------------------------ |
+| `SpawnOptions` fields             | 19     | ~8 (identity + dispatch) |
+| `RunOptions` fields               | 15     | ~9 (config only)         |
+| `startAgent()` lines              | ~130   | ~50                      |
+| Callback layers                   | 3      | 0 (direct subscription)  |
+| Live `ctx` references in queue    | 1      | 0 (snapshot)             |
+| Scattered status-transition sites | 6      | 1 (state machine)        |
+
+---
 
 ## Relationship with upstream
 

package/docs/plans/0102-consolidate-test-record-factory.md

@@ -0,0 +1,176 @@
+---
+issue: 102
+issue_title: "Consolidate test AgentRecord construction into a shared factory"
+---
+
+# Consolidate test AgentRecord construction into a shared factory
+
+## Problem Statement
+
+Eight test files independently construct `AgentRecord` objects using three different patterns: copy-pasted `makeRecord()`/`mockRecord()` factory functions (5 files), inline `const baseRecord: AgentRecord = { ... }` literals (2 files), and `as AgentRecord` casts.
+When issue #98 converts `AgentRecord` from an interface to a class, every object-literal construction site breaks.
+A shared factory confines that future breakage to a single file.
+
+## Goals
+
+- Create a shared `createTestRecord()` factory in `test/helpers/make-record.ts`.
+- Migrate all 7 affected test files to import the shared factory.
+- No production code changes.
+- No behavior changes — purely mechanical.
+
+## Non-Goals
+
+- Converting `AgentRecord` to a class — that is issue #98, which depends on this change.
+- Adding new test coverage — this is a refactoring of test infrastructure only.
+- Touching `test/agent-manager.test.ts` — it constructs records via `manager.spawn()`, not literals.
+- Consolidating other test helpers (mock sessions, mock TUI, etc.).
+
+## Background
+
+### Relevant modules
+
+| Module                               | Role                                                                                                           |
+| ------------------------------------ | -------------------------------------------------------------------------------------------------------------- |
+| `src/types.ts`                       | Defines the `AgentRecord` interface (20+ fields)                                                               |
+| `test/tools/agent-tool.test.ts`      | `makeRecord()` factory — 12 default fields, status "completed"                                                 |
+| `test/tools/get-result-tool.test.ts` | `makeRecord()` factory — 10 default fields, status "completed"                                                 |
+| `test/tools/steer-tool.test.ts`      | `makeRecord()` factory — 9 default fields, status "running", includes mock session, uses `as AgentRecord` cast |
+| `test/ui/agent-menu.test.ts`         | `makeRecord()` factory — 10 default fields, status "completed"                                                 |
+| `test/conversation-viewer.test.ts`   | `mockRecord()` factory — 6 default fields, status "running", uses `as AgentRecord` cast                        |
+| `test/notification.test.ts`          | 4 inline `baseRecord` literals, status "completed"                                                             |
+| `test/service-adapter.test.ts`       | 4 inline `baseRecord` / `minimal` literals, mixed statuses                                                     |
+
+### Convention from sibling packages
+
+`packages/pi-autoformat/test/helpers/rpc.ts` is the only existing shared test helper in the monorepo.
+The pattern is a plain module under `test/helpers/` with named exports — no class, no framework.
+
+### Relationship to issue #98
+
+Issue #98 plans to extract `MutableAgentRecord` as a class implementing the `AgentRecord` interface.
+That plan explicitly notes: "All test files that construct `AgentRecord` literals — they create interface-compatible objects, not class instances" and lists them as unchanged.
+Once this consolidation lands, issue #98's "unchanged" assumption becomes trivially true: only the shared factory needs updating if the construction API changes.
+
+## Design Overview
+
+### Shared factory: `createTestRecord()`
+
+A single function in `test/helpers/make-record.ts` with the `Partial<AgentRecord>` override pattern already used by 5 of the 7 files:
+
+```typescript
+import type { AgentRecord } from "../../src/types.js";
+
+export function createTestRecord(
+  overrides: Partial<AgentRecord> = {},
+): AgentRecord {
+  return {
+    id: "agent-1",
+    type: "general-purpose",
+    description: "Test task",
+    status: "completed",
+    result: "All done.",
+    toolUses: 3,
+    startedAt: 1000,
+    completedAt: 2000,
+    compactionCount: 0,
+    lifetimeUsage: { input: 500, output: 500, cacheWrite: 0 },
+    ...overrides,
+  };
+}
+```
+
+### Default-value decisions
+
+The defaults match the majority pattern (6 of 7 files default to a "completed" record).
+The two files that need "running" records (`steer-tool`, `conversation-viewer`) pass `{ status: "running" }` as overrides — a one-field change.
+
+The `as AgentRecord` cast used by `steer-tool.test.ts` and `conversation-viewer.test.ts` is no longer needed: the shared factory returns a full `AgentRecord` with all required fields populated, so TypeScript is satisfied without casting.
+
+### Migration strategy for inline-literal files
+
+`notification.test.ts` and `service-adapter.test.ts` construct multiple distinct inline literals — they don't have a single factory.
+Each inline literal becomes a `createTestRecord({ ...specific overrides })` call.
+The `baseRecord` variable declared in each `describe` block is replaced with a call to `createTestRecord()`.
+
+For `service-adapter.test.ts`, the top-level `baseRecord` with custom values (`id: "abc-123"`, `type: "Explore"`, etc.) becomes `createTestRecord({ id: "abc-123", type: "Explore", ... })`.
+
+## Module-Level Changes
+
+### New files
+
+1. `test/helpers/make-record.ts` — exports `createTestRecord()`.
+
+### Changed files
+
+1. `test/tools/agent-tool.test.ts` — remove local `makeRecord()`, import `createTestRecord` from helpers.
+2. `test/tools/get-result-tool.test.ts` — remove local `makeRecord()`, import `createTestRecord` from helpers.
+3. `test/tools/steer-tool.test.ts` — remove local `makeRecord()`, import `createTestRecord` from helpers.
+   Replace default `status: "running"` and `session` with overrides in each call site.
+4. `test/ui/agent-menu.test.ts` — remove local `makeRecord()`, import `createTestRecord` from helpers.
+5. `test/conversation-viewer.test.ts` — remove local `mockRecord()`, import `createTestRecord` from helpers.
+   Replace default `status: "running"` and `startedAt: Date.now()` with overrides in each call site.
+6. `test/notification.test.ts` — replace 4 inline `baseRecord` literals with `createTestRecord()` calls.
+7. `test/service-adapter.test.ts` — replace inline `baseRecord` / `minimal` / per-test literals with `createTestRecord()` calls.
+
+### Unchanged files
+
+1. `test/agent-manager.test.ts` — constructs records via `manager.spawn()`, not literals.
+2. All production source files — no changes.
+
+## Test Impact Analysis
+
+### New tests enabled
+
+1. A small sanity test in `test/helpers/make-record.test.ts` verifying that `createTestRecord()` returns a valid `AgentRecord` with expected defaults and that overrides are applied.
+   This is optional — the factory is exercised transitively by every consumer — but it documents the contract for future maintainers (especially when #98 changes construction).
+
+### Existing tests that become redundant
+
+None.
+This is a pure refactoring of test infrastructure; no production behavior changes.
+
+### Existing tests that stay as-is
+
+All existing test assertions stay unchanged.
+Only the construction of `AgentRecord` objects in test setup code changes; the assertions that read those records are untouched.
+
+## TDD Order
+
+1. **Create shared factory and its test.**
+   Add `test/helpers/make-record.ts` with `createTestRecord()`.
+   Add `test/helpers/make-record.test.ts` verifying defaults and override behavior.
+   Commit: `test: add shared createTestRecord factory (#102)`
+
+2. **Migrate tool test files.**
+   Update `agent-tool.test.ts`, `get-result-tool.test.ts`, `steer-tool.test.ts` to import `createTestRecord` and remove local `makeRecord()` functions.
+   Run `pnpm vitest run test/tools/agent-tool.test.ts test/tools/get-result-tool.test.ts test/tools/steer-tool.test.ts` to verify.
+   Commit: `test: migrate tool tests to shared createTestRecord (#102)`
+
+3. **Migrate UI test files.**
+   Update `agent-menu.test.ts` and `conversation-viewer.test.ts` to import `createTestRecord` and remove local `makeRecord()`/`mockRecord()` functions.
+   Run `pnpm vitest run test/ui/agent-menu.test.ts test/conversation-viewer.test.ts` to verify.
+   Commit: `test: migrate UI tests to shared createTestRecord (#102)`
+
+4. **Migrate notification and service-adapter tests.**
+   Update `notification.test.ts` and `service-adapter.test.ts` to replace inline literals with `createTestRecord()` calls.
+   Run `pnpm vitest run test/notification.test.ts test/service-adapter.test.ts` to verify.
+   Commit: `test: migrate notification and service-adapter tests to shared createTestRecord (#102)`
+
+5. **Final verification.**
+   Run full test suite (`pnpm vitest run`) and type check (`pnpm run check`) to confirm no regressions.
+   Commit: not needed if steps 2–4 are clean; otherwise a fix-up commit.
+
+## Risks and Mitigations
+
+| Risk                                                                                                                   | Mitigation                                                                                                                               |
+| ---------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- |
+| Shared defaults don't match a test's assumptions, causing silent false-passes                                          | Each migration step runs the affected test file immediately; review each test's overrides to ensure they still express the test's intent |
+| `steer-tool.test.ts` relies on `session: { fake: true }` in its factory default, which the shared factory omits        | Pass `session` as an override at each call site; the mock session is test-specific and doesn't belong in shared defaults                 |
+| `conversation-viewer.test.ts` uses `startedAt: Date.now()` which the shared factory replaces with `1000`               | Replace with `createTestRecord({ status: "running" })`; `startedAt` value is not asserted in any conversation-viewer test                |
+| `service-adapter.test.ts` uses custom `id`, `type`, `description` values that carry semantic meaning in its assertions | Pass those values explicitly as overrides to `createTestRecord()`                                                                        |
+| The `as AgentRecord` cast removal changes type-checking strictness                                                     | The shared factory returns a complete object satisfying all required fields, so removing the cast is strictly safer                      |
+
+## Open Questions
+
+- The factory name `createTestRecord` vs `makeRecord` vs `makeAgentRecord`: the plan uses `createTestRecord` to distinguish it from the production `AgentRecord` constructor that #98 will introduce.
+  If #98 names its constructor differently, this can be revisited.

package/docs/retro/0061-session-format-transcript.md

@@ -0,0 +1,41 @@
+---
+issue: 61
+issue_title: "feat: port subagent transcript logging to Pi's official JSONL session format"
+---
+
+# Retro: #61 — port subagent transcript logging to Pi's official JSONL session format
+
+## Final Retrospective (2026-05-20T17:15:00Z)
+
+### Session summary
+
+Planned, implemented, and shipped a migration from the bespoke `output-file.ts` transcript format to Pi's official JSONL session format via `SessionManager.create()`.
+The change replaced 143 lines of manual streaming code with 3 lines leveraging the SDK's native persistence, nested subagent sessions under the parent session directory with `parentSession` header linking.
+Released as `pi-subagents-v6.0.0` (major version bump due to breaking transcript format change).
+
+### Observations
+
+#### What went well
+
+- The plan-to-implementation translation was clean: 6 TDD steps mapped to 7 commits (one extra `fix:` for biome lint).
+  No steps needed reordering or merging.
+- The `ask_user` design decision gates during planning (persistence strategy, file location) produced clear answers that avoided rework during implementation.
+- Research into nicobailon/pi-subagents, edxeth/pi-subagents, and HazAT/pi-interactive-subagents provided useful reference for the session directory layout, confirming the parent-relative nesting pattern.
+- The biome lint catch on the unused `cwd` parameter led to a better design — incorporating `cwd` into the temp fallback path for project namespacing — rather than a mechanical underscore prefix.
+
+#### What caused friction (agent side)
+
+- `missing-context` — The plan listed test impact for `agent-runner.test.ts` but didn't grep for other test files mocking `SessionManager` or `ctx.sessionManager`.
+  Three additional files needed updating: `agent-runner-extension-tools.test.ts`, `print-mode.test.ts`, and `test/tools/agent-tool.test.ts`.
+  The testing skill explicitly says "grep for ALL test files that construct a compatible mock — not just factory helpers."
+  Impact: ~5 minutes of reactive fixes during Step 4.
+  Self-identified at implementation time.
+
+- `missing-context` — The plan didn't account for the timing difference between the old synchronous `record.outputFile` assignment (immediately after `spawn()`) and the new asynchronous availability (after `SessionManager.create()` runs inside `runAgent()`).
+  This required adding `session.sessionManager.getSessionFile()` in the `onSessionCreated` callback — a design decision made during implementation.
+  Impact: minor within-step rework, no extra commit needed.
+
+#### What caused friction (user side)
+
+- The dependency update to 0.75.4 was a reasonable pre-plan request, but it added ~10 minutes of tangential work (diagnosing `pnpm update` resolution behavior, normalizing version specifiers).
+  This could have been a separate commit/session, though batching it was pragmatic since it gave the plan access to the latest SDK types.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@gotgenes/pi-subagents",
-  "version": "6.0.0",
+  "version": "6.0.1",
   "exports": {
     ".": "./src/service.ts"
   },