@falai/agent 1.2.7 → 2.0.0

package/docs/migration/v1-to-v2.md

@@ -0,0 +1,909 @@
+---
+title: "v1 → v2 migration"
+description: "Every breaking change in v2 with rename tables and before/after."
+type: migration
+order: 1
+---
+
+# v1 → v2 Migration
+
+**Version:** 2.0.0 — Clean break (no shims, no aliases)
+
+## Summary
+
+v2 is a clean break — no shims, no aliases. The surface is smaller and the primitives are fewer: `Route` becomes `Flow`, three behavioral types collapse into `Instruction`, `EnhancedTool` folds into `Tool`, and all flow control converges on a single `Directive` shape. This guide covers every breaking change with rename tables and before/after code samples.
+
+Other notable changes covered below: the `pendingTransition` session field becomes `pendingDirective`, `Tool.name` collapses into `Tool.id`, the `ConditionTemplate` union is replaced by separate `when` (AI) and `if` (code) fields, `FlowOptions.onComplete` is string-only at the top level, `createAgent` becomes the headline API, agent identity collapses from three fields (`description`, `identity`, `personality`) into one (`persona`), multi-step batching is replaced by `auto: true` steps, and `step.branches` is added as an explicit (non-breaking) alternative to implicit fork.
+
+---
+
+## Table of Contents
+
+1. [Persistence: `pendingTransition` → `pendingDirective`](#1-persistence-pendingtransition--pendingdirective)
+2. [Removed in v2 — Full Surface List](#2-removed-in-v2--full-surface-list)
+3. [Route → Flow Rename](#3-route--flow-rename)
+4. [Guideline / Rule / Prohibition → Instruction](#4-guideline--rule--prohibition--instruction)
+5. [`condition` / `action` → `when` / `prompt`](#5-condition--action--when--prompt)
+6. [`onComplete` Cleanup](#6-oncomplete-cleanup)
+7. [Flow Completion: Farewell Removal + Idle-State Release](#7-flow-completion-farewell-removal--idle-state-release)
+8. [`createAgent` — The New Headline API](#8-createagent--the-new-headline-api)
+9. [Tool / EnhancedTool Merge](#9-tool--enhancedtool-merge)
+10. [ConditionTemplate → `when` / `if` Split](#10-conditiontemplate--when--if-split)
+11. [Dispatch Replaces `transitionTo` / `nextStepFlow`](#11-dispatch-replaces-transitionto--nextstepflow)
+12. [Agent Identity → `persona`](#12-agent-identity--persona)
+13. [Multi-Step Batching → Auto-Steps](#13-multi-step-batching--auto-steps)
+14. [Implicit Fork → Explicit Branches (Non-Breaking)](#14-implicit-fork--explicit-branches-non-breaking)
+
+---
+
+## 1. Persistence: `pendingTransition` → `pendingDirective`
+
+The `SessionState.pendingTransition` field is removed. Its replacement is `SessionState.pendingDirective`, a full `Directive` object (not just a flow id + reason).
+
+v2 also adds a `signals` column for forward-compat with the v2.x Signals feature. It's unused at runtime but must be present in your schema.
+
+**There is no auto-migration.** If a v1 record carries `pendingTransition` but no `pendingDirective`, v2 ignores it — the pending transition is lost. Run the backfill before deploying v2.
+
+### PostgreSQL / SQLite
+
+```sql
+-- Step 1: Backfill pendingDirective from pendingTransition
+UPDATE sessions
+SET pending_directive = jsonb_build_object(
+  'goTo', jsonb_build_object('flow', pending_transition->'targetFlowId')
+)
+WHERE pending_transition IS NOT NULL
+  AND pending_directive IS NULL;
+
+-- Step 2: Schema migration
+ALTER TABLE sessions DROP COLUMN pending_transition;
+ALTER TABLE sessions ADD COLUMN pending_directive JSONB;
+ALTER TABLE sessions ADD COLUMN signals JSONB;
+```
+
+For SQLite (no JSONB), use TEXT columns and `json_object()`:
+
+```sql
+UPDATE sessions
+SET pending_directive = json_object(
+  'goTo', json_object('flow', json_extract(pending_transition, '$.targetFlowId'))
+)
+WHERE pending_transition IS NOT NULL
+  AND pending_directive IS NULL;
+
+ALTER TABLE sessions DROP COLUMN pending_transition;
+ALTER TABLE sessions ADD COLUMN pending_directive TEXT;
+ALTER TABLE sessions ADD COLUMN signals TEXT;
+```
+
+### MongoDB
+
+```javascript
+db.sessions.updateMany(
+  { pendingTransition: { $ne: null }, pendingDirective: { $exists: false } },
+  [{
+    $set: {
+      pendingDirective: {
+        goTo: { flow: "$pendingTransition.targetFlowId" }
+      }
+    }
+  }]
+);
+
+// Drop legacy field
+db.sessions.updateMany({}, { $unset: { pendingTransition: "" } });
+```
+
+### Redis
+
+Extend your migration Lua script (see the [route-to-flow migration](./route-to-flow.md) for the pattern):
+
+```lua
+local cursor = "0"
+repeat
+  local result = redis.call("SCAN", cursor, "MATCH", "session:*", "COUNT", 100)
+  cursor = result[1]
+  for _, key in ipairs(result[2]) do
+    local val = redis.call("GET", key)
+    if val then
+      -- Replace pendingTransition with pendingDirective in JSON
+      val = val:gsub('"pendingTransition"', '"pendingDirective"')
+      redis.call("SET", key, val)
+    end
+  end
+until cursor == "0"
+```
+
+> **Note:** Redis stores serialized JSON. The field-name swap is sufficient if your `pendingTransition` shape was `{ targetFlowId: string }`. The v2 runtime expects `{ goTo: { flow: string } }` — adjust the transform if your v1 shape was different.
+
+### OpenSearch
+
+Use the `_reindex` API with a painless script:
+
+```json
+POST _reindex
+{
+  "source": { "index": "sessions" },
+  "dest": { "index": "sessions_v2" },
+  "script": {
+    "source": """
+      if (ctx._source.containsKey('pendingTransition') && ctx._source.pendingTransition != null) {
+        def pt = ctx._source.remove('pendingTransition');
+        ctx._source.pendingDirective = ['goTo': ['flow': pt.targetFlowId]];
+      }
+    """,
+    "lang": "painless"
+  }
+}
+```
+
+Then swap the alias to `sessions_v2`.
+
+### Prisma Schema
+
+```diff
+model Session {
+  // ...
+- pendingTransition Json?  @map("pending_transition")
++ pendingDirective  Json?  @map("pending_directive")
++ signals           Json?
+}
+```
+
+Run `npx prisma migrate dev --name v2-directive-migration`.
+
+---
+
+## 2. Removed in v2 — Full Surface List
+
+v2 is a clean break. These symbols are gone with no aliases:
+
+| v1 Element | Replacement | Migration |
+|---|---|---|
+| `END_FLOW` / `END_FLOW_ID` | Implicit terminus + `{ complete: true }` | Remove sentinel references. Last step in `steps[]` auto-terminates. |
+| `ConditionTemplate` | `when` (AI) + `if` (code) | See [§10](#10-conditiontemplate--when--if-split). |
+| `EnhancedTool` | `Tool` (merged) | See [§9](#9-tool--enhancedtool-merge). |
+| `FlowTransitionConfig` | `Directive` | `{ nextStep: 'X', condition: 'Y' }` → `{ goTo: 'X', reason: 'Y' }` |
+| `FlowCompletionHandler` | `hooks.onComplete: (ctx) => Directive \| void` | Inline the signature change. |
+| `FlowOptions.endStep` | Last step's `prompt` | Move closing text to the last step. |
+| `Flow.endStepSpec` | Removed | Internal — gone with `endStep`. |
+| `Step.endFlow()` | Implicit terminus | `.endFlow(opts)` → just make it the last step. |
+| `Agent.nextStepFlow()` | `Agent.dispatch()` | See [§11](#11-dispatch-replaces-transitionto--nextstepflow). |
+| `Agent.transitionTo()` | `Agent.dispatch()` | See [§11](#11-dispatch-replaces-transitionto--nextstepflow). |
+| `PendingTransition` interface | `Directive` | Replace with `pendingDirective: Directive`. |
+| `SessionState.pendingTransition` | `SessionState.pendingDirective` | See [§1](#1-persistence-pendingtransition--pendingdirective). |
+| `StepResult` (export) | Internal only | Chain methods stay; the return type is no longer public. |
+| `BranchResult` / `BranchSpec` | `BranchEntry` / `BranchMap` | Use `step.branches: [{ if?, when?, then, label? }]`. |
+| `ResponseModal` (export) | Internal | Remove any direct imports. |
+| `ResponsePipeline` (export) | Internal | Remove any direct imports. |
+| `BatchExecutor` / `BatchPromptBuilder` | Auto-steps | See [§13](#13-multi-step-batching--auto-steps). |
+| `PromptComposer` (export) | Internal | Remove any direct imports. |
+| `PromptSectionCache` (export) | Internal | Remove any direct imports. |
+| `CompactionEngine` (export) | Internal | Remove any direct imports. |
+| `FlowRouter` (export) | Internal | Remove any direct imports. |
+| `AutoChainExecutor` (export) | Internal | Remove any direct imports. |
+| `Guideline` / `Rule` / `Prohibition` | `Instruction` | See [§4](#4-guideline--rule--prohibition--instruction). |
+| `Tool.name` | `Tool.id` | The LLM sees `tool.id` as the tool name. See [§9](#9-tool--enhancedtool-merge). |
+| `AgentOptions.description` | `AgentOptions.persona` | See [§12](#12-agent-identity--persona). |
+| `AgentOptions.identity` | `AgentOptions.persona` | See [§12](#12-agent-identity--persona). |
+| `AgentOptions.personality` | `AgentOptions.persona` | See [§12](#12-agent-identity--persona). |
+| `AgentOptions.compositionMode` | Removed | Had no runtime effect. Delete any references. |
+| `AgentOptions.maxStepsPerBatch` | `AgentOptions.maxAutoStepsPerTurn` | See [§13](#13-multi-step-batching--auto-steps). |
+| `StepOptions.skipIf` | `StepOptions.skip` | Function-only field. See [§10](#10-conditiontemplate--when--if-split). |
+| `StepOptions.step` | Removed | Dead code since the `END_FLOW` symbol removal. |
+| `FlowOptions.terms` / `FlowOptions.knowledgeBase` | Agent-level only | Move to `AgentOptions`. |
+| `FlowOptions.identity` / `FlowOptions.personality` | `AgentOptions.persona` or flow-level instruction | See [§12](#12-agent-identity--persona). |
+| `appliedGuidelines` (response) | `appliedInstructions` | Rename in response handling code. |
+| `StoppedReason: 'end_flow'` | `'last_step'` | Update assertions. |
+| `StoppedReason: 'flow_complete'` | `'last_step'` or `'completed'` | Update assertions. |
+| `StoppedReason: 'max_steps_reached'` | `'max_auto_steps'` | Update assertions. |
+
+### New `StoppedReason` values
+
+```typescript
+type StoppedReason =
+  | 'needs_input'       // unchanged
+  | 'last_step'         // last step ran — no successor
+  | 'completed'         // explicit { complete: true } directive
+  | 'aborted'           // { abort: '...' } directive
+  | 'goto'              // goTo/goToStep directive short-circuited
+  | 'reset'             // { reset: true } directive
+  | 'halt'              // PreDirective halt: true
+  | 'reply'             // Step.reply or directive reply (LLM skipped)
+  | 'max_auto_steps'    // auto-step chain cap hit
+  | 'prepare_error'     // unchanged
+  | 'llm_error'         // unchanged
+  | 'validation_error'  // unchanged
+  | 'finalize_error';   // unchanged
+```
+
+---
+
+## 3. Route → Flow Rename
+
+The `Route` domain noun was renamed to `Flow` in a minor breaking bump that ships before v2. v2 assumes `Flow` naming everywhere — `Route` / `createRoute` / `session.currentRoute` and the corresponding persistence columns no longer exist.
+
+If you haven't run that migration yet, start there: see **[Route → Flow rename](./route-to-flow.md)** for the full rename table, schema changes per adapter, and Redis/OpenSearch backfill scripts. Once that lands, return here for the rest of the v2 changes.
+
+
+---
+
+## 4. Guideline / Rule / Prohibition → Instruction
+
+The three behavioral primitives collapse into a single `Instruction` type with a `kind` discriminator:
+
+### Rename Table
+
+| v1 | v2 `Instruction.kind` | Notes |
+|---|---|---|
+| `Rule` (always do) | `kind: 'must'` | Rendered as `[must]` prefix |
+| `Prohibition` (never do) | `kind: 'never'` | Rendered as `[never]` prefix |
+| `Guideline` (should do) | `kind: 'should'` | Default when `kind` omitted |
+
+### Before / After
+
+```typescript
+// ─── v1 ───
+agent.addRule("Always greet the user by name");
+agent.addProhibition("Never discuss competitors");
+agent.addGuideline({
+  when: "User is a returning customer",
+  action: "Skip the introduction and get to the point",
+});
+
+// ─── v2 ───
+agent.addInstruction({ kind: 'must', prompt: "Always greet the user by name" });
+agent.addInstruction({ kind: 'never', prompt: "Never discuss competitors" });
+agent.addInstruction({
+  kind: 'should',
+  when: "User is a returning customer",
+  prompt: "Skip the introduction and get to the point",
+});
+```
+
+### Top-level options
+
+```typescript
+// Before
+const agent = createAgent({
+  rules: ['Always verify email format'],
+  prohibitions: ['Never promise delivery dates'],
+  guidelines: [{ when: 'User is upset', prompt: 'Use empathetic tone' }],
+});
+
+// After
+const agent = createAgent({
+  instructions: [
+    { kind: 'must', prompt: 'Always verify email format' },
+    { kind: 'never', prompt: 'Never promise delivery dates' },
+    { kind: 'should', when: 'User is upset', prompt: 'Use empathetic tone' },
+  ],
+});
+```
+
+The same applies at flow and step scope. `FlowOptions.rules`, `FlowOptions.prohibitions`, `FlowOptions.guidelines`, and `StepOptions.guidelines` are all removed — use `instructions` at each level.
+
+### Prompt Rendering Change
+
+The heading `## Behavioral Guidelines` becomes `## Instructions`. Line format:
+
+```
+[must] [Always] Always greet the user by name
+[never] [Always] Never discuss competitors
+[should] [When: User is a returning customer] Skip the introduction and get to the point
+```
+
+### Response Field
+
+`response.appliedGuidelines` → `response.appliedInstructions`
+
+```typescript
+// v1
+console.log(response.appliedGuidelines);
+
+// v2
+console.log(response.appliedInstructions);
+```
+
+### Deprecated method removal (Instructions)
+
+| Old | New |
+|-----|-----|
+| `agent.createGuideline(g)` | `agent.createInstruction(g)` |
+| `agent.getRules()` | `agent.instructions.filter(i => i.kind === 'must')` |
+| `agent.getProhibitions()` | `agent.instructions.filter(i => i.kind === 'never')` |
+| `flow.createGuideline(g)` | `flow.createInstruction(g)` |
+| `flow.getGuidelines()` | `flow.getInstructions()` |
+| `flow.getRules()` | Removed — use flow instructions |
+| `flow.getProhibitions()` | Removed — use flow instructions |
+| `step.addGuideline(g)` | `step.addInstruction(g)` |
+| `step.getGuidelines()` | `step.getInstructions()` |
+
+### Type alias removal
+
+| Old type | Use instead |
+|----------|-------------|
+| `Guideline<C, D>` | `Instruction<C, D>` |
+| `ScopedGuidelines<C, D>` | `ScopedInstructions<C, D>` |
+| `AppliedGuideline` | `AppliedInstruction` |
+
+---
+
+## 5. `condition` / `action` → `when` / `prompt`
+
+The guideline shape fields were renamed in v2:
+
+| v1 Field | v2 Field | Purpose |
+|---|---|---|
+| `condition` | `when` | AI-evaluated activation condition |
+| `action` | `prompt` | The behavioral instruction text |
+
+### Before / After
+
+```typescript
+// ─── v1 ───
+agent.addGuideline({
+  condition: "User is frustrated",
+  action: "Be extra empathetic and offer to escalate",
+});
+
+// ─── v2 ───
+agent.addInstruction({
+  when: "User is frustrated",
+  prompt: "Be extra empathetic and offer to escalate",
+});
+```
+
+This also applies to flow-scoped and step-scoped guidelines:
+
+```typescript
+// ─── v1 ───
+flow.addGuideline({
+  condition: "Discussing pricing",
+  action: "Always mention the free tier first",
+});
+
+// ─── v2 ───
+flow.addInstruction({
+  when: "Discussing pricing",
+  prompt: "Always mention the free tier first",
+});
+```
+
+---
+
+## 6. `onComplete` Cleanup
+
+### What changed
+
+- **Top-level** `FlowOptions.onComplete` accepts **only a string** (flow id sugar).
+- **Function** and **Directive** forms move to `hooks.onComplete`.
+- Setting both throws `FlowConfigurationError` at construction.
+
+### Before / After
+
+```typescript
+// ─── v1: function at top level ───
+agent.createFlow({
+  title: "Booking",
+  onComplete: (session, context) => {
+    return { nextRoute: "Confirmation" };
+  },
+  steps: [/* ... */],
+});
+
+// ─── v2: string at top level (sugar) ───
+agent.createFlow({
+  title: "Booking",
+  onComplete: "Confirmation",  // sugar for hooks.onComplete = () => ({ goTo: 'Confirmation' })
+  steps: [/* ... */],
+});
+
+// ─── v2: function in hooks ───
+agent.createFlow({
+  title: "Booking",
+  hooks: {
+    onComplete: (ctx) => {
+      if (ctx.data.isPremium) return { goTo: "PremiumConfirmation" };
+      return { goTo: "Confirmation" };
+    },
+  },
+  steps: [/* ... */],
+});
+```
+
+### ⚠️ Do NOT set both
+
+```typescript
+// THROWS FlowConfigurationError at construction
+agent.createFlow({
+  title: "Booking",
+  onComplete: "Confirmation",          // ← string sugar
+  hooks: { onComplete: () => {...} },   // ← function form — conflict!
+});
+```
+
+---
+
+## 7. Flow Completion: Farewell Removal + Idle-State Release
+
+### Hardcoded farewell removed
+
+v1 injected a synthetic `__COMPLETED__` step with a hardcoded prompt (`"Send a brief, natural farewell message…"`) when a flow completed. This is **gone**. No framework-generated farewell message is emitted. Every word the user sees comes from your step prompts.
+
+### Idle-state release
+
+When a flow completes and `onComplete` does not produce a transition:
+
+- `session.currentFlow` → `undefined`
+- `session.currentStep` → `undefined`
+- The flow is marked `completed: true` in `flowHistory`
+- The router excludes completed flows from future scoring
+- Next turn: routing runs fresh (or the no-flow fallback triggers)
+
+### Migration: Add an explicit closing step
+
+```typescript
+// ─── v1: relied on framework-generated farewell ───
+agent.createFlow({
+  title: "Onboarding",
+  steps: [
+    { id: "name",  collect: ["name"]  },
+    { id: "email", collect: ["email"] },
+  ],
+});
+// Framework would auto-generate "Thank you! I've recorded all..."
+
+// ─── v2: author your own closing turn ───
+agent.createFlow({
+  title: "Onboarding",
+  steps: [
+    { id: "name",  collect: ["name"]  },
+    { id: "email", collect: ["email"] },
+    { id: "thanks", prompt: "Thank the user warmly. Wish them a great day." },
+  ],
+});
+```
+
+### `flow.reentrant` opt-in
+
+If your v1 code relied on the router re-entering a completed flow, that loop is gone. To restore it deliberately:
+
+```typescript
+agent.createFlow({
+  title: "Search",
+  reentrant: true,  // allows re-selection after completion
+  requiredFields: ["query"],
+  steps: [/* ... */],
+});
+```
+
+When `reentrant: true`, the router can re-select this flow after it completes. On re-entry, fields declared in `requiredFields` / `optionalFields` are cleared so the flow starts fresh.
+
+`onComplete` always wins over `reentrant` — if `onComplete` returns a target, the session goes there instead.
+
+---
+
+## 8. `createAgent` — The New Headline API
+
+`createAgent` is the recommended entry point in v2. It's equivalent to `new Agent(options)` but reads better in examples and enables stronger generic inference.
+
+```typescript
+import { createAgent, GeminiProvider } from "@falai/agent";
+
+const agent = createAgent({
+  name: "BookingBot",
+  provider: new GeminiProvider({
+    apiKey: process.env.GEMINI_API_KEY!,
+    model: "models/gemini-2.5-flash",
+  }),
+  schema: {
+    type: "object",
+    properties: {
+      destination: { type: "string" },
+      date: { type: "string" },
+      guests: { type: "number" },
+    },
+    required: ["destination", "date", "guests"],
+  },
+  flows: [
+    {
+      title: "Book Trip",
+      when: ["User wants to book travel"],
+      requiredFields: ["destination", "date", "guests"],
+      steps: [
+        { id: "ask_dest", prompt: "Where to?", collect: ["destination"] },
+        { id: "ask_date", prompt: "When?", collect: ["date"] },
+        { id: "ask_guests", prompt: "How many?", collect: ["guests"] },
+      ],
+    },
+  ],
+});
+
+const response = await agent.respond("Book a trip to Paris for 2 next Friday");
+```
+
+`new Agent(options)` still works — `createAgent` is sugar, not a replacement.
+
+
+---
+
+## 9. Tool / EnhancedTool Merge
+
+`EnhancedTool` is removed. Its optional metadata fields are now part of the base `Tool` interface. `Tool.name` is also removed — `Tool.id` is the sole identifier and is what the LLM sees as the tool name.
+
+```typescript
+// ─── v1 ───
+import { EnhancedTool } from "@falai/agent";
+
+const tool: EnhancedTool = {
+  id: "search",
+  name: "Search Database",
+  description: "Search the database",
+  parameters: { type: "object", properties: {} },
+  handler: async (ctx, args) => { /* ... */ },
+  isConcurrencySafe: true,
+  isReadOnly: true,
+  checkPermissions: async (ctx) => ({ allowed: true }),
+};
+
+// ─── v2 ───
+import { Tool } from "@falai/agent"; // EnhancedTool is gone
+
+const tool: Tool = {
+  id: "search_database",            // Tool.id is the sole identifier
+  description: "Search the database",
+  parameters: { type: "object", properties: {} },
+  handler: async (ctx, args) => { /* ... */ },
+  // Optional metadata (formerly EnhancedTool-only):
+  isConcurrencySafe: true,
+  isReadOnly: true,
+  checkPermissions: async (ctx) => ({ allowed: true }),
+};
+```
+
+Choose descriptive IDs — the LLM sees them. The optional fields that moved to `Tool`: `isConcurrencySafe`, `isReadOnly`, `isDestructive`, `interruptBehavior`, `maxResultSizeChars`, `validateInput`, `checkPermissions`.
+
+### New: `ToolContext.dispatch` and `ToolResult.directive`
+
+Tools can now control flow directly:
+
+```typescript
+const refundTool: Tool = {
+  id: "check_refund",
+  description: "Check refund eligibility",
+  parameters: { /* ... */ },
+  handler: async (ctx, args) => {
+    const order = await lookupOrder(args.orderId);
+
+    if (order.age > 90) {
+      // Imperative: redirect mid-handler
+      ctx.dispatch({ goTo: "IneligibleRefund" });
+      return { data: "Order too old for refund" };
+    }
+
+    // Declarative: return directive alongside result
+    return {
+      data: "Eligible for refund",
+      directive: { goTo: "ProcessRefund" },
+    };
+  },
+};
+```
+
+### Deprecated method removal (Tools)
+
+| Old | New |
+|-----|-----|
+| `agent.createTool(t)` | `agent.addTool(t)` |
+
+### Type alias removal (Tools)
+
+| Old type | Use instead |
+|----------|-------------|
+| `EnhancedTool<C, D, R>` | `Tool<C, D, R>` |
+
+---
+
+## 10. ConditionTemplate → `when` / `if` Split
+
+The v1 `ConditionTemplate` type (a union of `string | function | Array<string | function>`) is removed. Conditions are now split into two distinct fields:
+
+| Field | Evaluator | Cost | Accepts |
+|---|---|---|---|
+| `when` | AI (LLM call) | Tokens | `string \| string[]` |
+| `if` | Code (function) | Free | `(ctx) => boolean \| Array<(ctx) => boolean>` |
+
+### Rules
+
+- `when` accepts **only strings** — passing a function throws `FlowConfigurationError` at construction.
+- `if` accepts **only functions**.
+- When both are set: `if` runs first (free); `when` runs only if `if` passes (saves tokens).
+- Arrays are AND-conjunctions.
+
+### Before / After
+
+```typescript
+// ─── v1: mixed ConditionTemplate ───
+agent.createFlow({
+  title: "Premium Support",
+  when: [
+    "User needs urgent help",                    // AI condition
+    (ctx) => ctx.context.userTier === "premium", // code condition
+  ],
+  steps: [/* ... */],
+});
+
+// ─── v2: split into when + if ───
+agent.createFlow({
+  title: "Premium Support",
+  when: ["User needs urgent help"],              // AI only
+  if: [(ctx) => ctx.context.userTier === "premium"], // code only
+  steps: [/* ... */],
+});
+```
+
+### `skipIf` → `skip`
+
+On steps, `skipIf` is renamed to `skip` and accepts only `if`-style functions:
+
+```typescript
+// ─── v1 ───
+{ id: "ask_name", skipIf: (data) => !!data.name, /* ... */ }
+
+// ─── v2 ───
+{ id: "ask_name", skip: (data) => !!data.name, /* ... */ }
+```
+
+---
+
+## 11. Dispatch Replaces `transitionTo` / `nextStepFlow`
+
+Both `Agent.transitionTo()` and `Agent.nextStepFlow()` are removed. Use `Agent.dispatch()`:
+
+```typescript
+// ─── v1 ───
+await agent.nextStepFlow("Feedback", session, "user requested feedback");
+await agent.transitionTo("Billing");
+
+// ─── v2 ───
+await agent.dispatch({ goTo: "Feedback", reason: "user requested feedback" }, session);
+await agent.dispatch("Billing"); // string sugar for { goTo: "Billing" }
+```
+
+`dispatch` sets `session.pendingDirective` — the directive is applied at the start of the next turn (not immediately). For synchronous in-place application without a `respond()` call, use `agent.applyDirective(directive, session)`.
+
+### Deprecated method removal (Dispatch & session access)
+
+| Old | New |
+|-----|-----|
+| `agent.transitionTo(...)` | `agent.dispatch({ goTo: ... }, session)` |
+| `agent.nextStepFlow(...)` | `agent.dispatch({ goTo: ... }, session)` |
+| `agent.getCurrentSession()` | `agent.currentSession` |
+| `agent.setCurrentSession(s)` | `agent.currentSession = s` |
+| `agent.clearCurrentSession()` | `agent.currentSession = undefined` |
+| `agent.getSchema()` | `agent.schema` |
+| `agent.getKnowledgeBase()` | `agent.knowledgeBase` |
+
+### Type alias removal (Dispatch)
+
+| Old type | Use instead |
+|----------|-------------|
+| `FlowTransitionConfig` | `Directive` |
+| `FlowCompletionHandler` | `hooks.onComplete` function |
+| `PendingTransition` | `Directive` (assigned to `session.pendingDirective`) |
+
+---
+
+## 12. Agent Identity → `persona`
+
+Three v1 agent fields — `description`, `identity`, `personality` — collapse into a single `persona` field. `persona` is a `Template<TContext>` covering role, tone, and self-concept. Merge your old copy into one coherent prompt.
+
+### Before / After
+
+```typescript
+// ─── v1 ───
+const agent = createAgent({
+  name: 'Support Bot',
+  description: 'A helpful customer support agent',
+  identity: 'You are a senior support specialist at Acme Corp.',
+  personality: 'Friendly, concise, solution-oriented',
+  // ...
+});
+
+// ─── v2 ───
+const agent = createAgent({
+  name: 'Support Bot',
+  goal: 'Help customers resolve issues quickly',
+  persona: 'You are a senior support specialist at Acme Corp. Communicate in a friendly, concise, solution-oriented style.',
+  // ...
+});
+```
+
+### Flow-level identity / personality removed
+
+`FlowOptions.identity` and `FlowOptions.personality` are removed. Use the agent-level `persona` for global voice, or a flow-level instruction for flow-specific voice:
+
+```typescript
+// Before
+const flow: FlowOptions = {
+  title: 'Billing',
+  identity: 'You are a billing specialist',
+  personality: 'Formal and precise',
+};
+
+// After — use a flow-level instruction
+const flow: FlowOptions = {
+  title: 'Billing',
+  instructions: [
+    { kind: 'should', prompt: 'Act as a billing specialist. Use formal, precise language.' },
+  ],
+};
+```
+
+### Flow-level terms / knowledgeBase removed
+
+Terms and knowledge base are now agent-level only. Move them up:
+
+```typescript
+// Before
+const flow: FlowOptions = {
+  title: 'Booking',
+  terms: [{ name: 'PNR', description: 'Passenger Name Record' }],
+  knowledgeBase: { policies: { /* ... */ } },
+};
+
+// After — move to agent level
+const agent = createAgent({
+  terms: [{ name: 'PNR', description: 'Passenger Name Record' }],
+  knowledgeBase: { policies: { /* ... */ } },
+  flows: [{ title: 'Booking', /* ... */ }],
+});
+```
+
+
+---
+
+## 13. Multi-Step Batching → Auto-Steps
+
+Multi-step batching (`maxStepsPerBatch`, `BatchExecutor`, `BatchPromptBuilder`) conflated two different concerns: skipping the LLM for non-interactive nodes, and compressing N user-facing steps into one response. v2 splits them: the first becomes `auto: true`, the second becomes a single step with multi-field `collect`.
+
+| Old | New |
+|-----|-----|
+| `AgentOptions.maxStepsPerBatch` | `AgentOptions.maxAutoStepsPerTurn` (default `10`) |
+| `BatchExecutor`, `BatchPromptBuilder`, `needsInput` | Removed |
+| Batch events (`batch_start`, `step_included`, `batch_complete`, …) | Step events (`step_entered`, `step_skipped`, `step_completed`) — `auto: boolean` in payload |
+| `StoppedReason: 'max_steps_reached'` | `'max_auto_steps'` |
+
+**Restructure pattern:** N tiny ask-steps batched into one call → one step with `collect: [field1, field2, field3]` and a prompt that asks for whatever is still missing. Pre-extraction handles the "user dumped everything in one message" case.
+
+**Auto-step pattern:** Computation between asks → mark the compute step `auto: true`. It runs `onEnter` / `prepare` / `branches` / `onExit` with no LLM call. The pipeline walks consecutive auto-steps until it hits an interactive step or terminating directive.
+
+```typescript
+// v2: enrichment between two asks
+{ id: 'ask_email', collect: ['email'], prompt: 'What email?' },
+{ id: 'enrich', auto: true, prepare: async ({ data }) => {
+    const customer = await crm.findByEmail(data.email);
+    return { contextUpdate: { isReturning: !!customer } };
+  } },
+{ id: 'confirm', prompt: ({ context }) =>
+    context.isReturning ? 'Welcome back. Confirm?' : 'Confirm your order?' },
+```
+
+**Validation:** an `auto: true` step throws `FlowConfigurationError` if it sets `prompt`, `collect`, `tools`, or `finalize`. `onEnter`, `prepare`, `onExit`, `branches`, `requires`, `skip` are all allowed. No schema change — auto-steps does not touch `SessionState`, existing persistence adapters need no migration.
+
+---
+
+## 14. Implicit Fork → Explicit Branches (Non-Breaking)
+
+This is **optional**. The implicit-fork pattern (multiple successor steps each with their own `step.when`) still works in v2. `step.branches` is a new declarative form for the same routing behavior, recommended when the fork is the point of the step rather than incidental to it.
+
+### When to convert
+
+- The source step exists solely to route — it's a decision point, not a conversation node.
+- Three or more successors with `when` conditions where the decision tree is hard to follow.
+- You want code-only routing (`if`) to skip LLM evaluation entirely.
+- You need mixed targets: some branches go to local steps, others jump to other flows or emit full Directives.
+
+### Stay implicit when
+
+- The flow reads as a linear chain A → B → C where B is occasionally skipped.
+- Only two successors, with self-explanatory `when` conditions.
+
+### Before / After
+
+```typescript
+// Before — routing scattered across 4 target steps
+{
+  steps: [
+    { id: 'triage', prompt: 'How can I help?' },
+    { id: 'billing',      when: 'asking about billing',       prompt: '…' },
+    { id: 'tech_support', when: 'asking a technical question', prompt: '…' },
+    { id: 'cancellation', when: 'wants to cancel',             prompt: '…' },
+    { id: 'general',      prompt: '…' },
+  ],
+}
+
+// After — routing declared once at the source
+{
+  steps: [
+    {
+      id: 'triage',
+      prompt: 'How can I help?',
+      branches: [
+        { when: 'asking about billing',        then: 'billing' },
+        { when: 'asking a technical question', then: 'tech_support' },
+        { when: 'wants to cancel',             then: 'cancellation' },
+        { then: 'general' }, // unconditional fallback (must be last)
+      ],
+    },
+    { id: 'billing',      prompt: '…' },
+    { id: 'tech_support', prompt: '…' },
+    { id: 'cancellation', prompt: '…' },
+    { id: 'general',      prompt: '…' },
+  ],
+}
+```
+
+Runtime behavior is identical: AI evaluates entries in declaration order; first match wins; an entry without `when`/`if` is the fallback. The only difference is where the routing logic lives.
+
+### Mixed targets
+
+Branches can route to local step ids, flow ids (sugar for `goTo`), or full Directives — implicit forks can only target steps in the same flow:
+
+```typescript
+branches: [
+  { if: ({ data }) => data.plan === 'enterprise', then: 'enterprise_path' },           // local step
+  { when: 'wants to cancel', then: 'CancellationFlow' },                                // flow id
+  { when: 'needs a refund', then: { goTo: { flow: 'Refund', data: { source: 'triage' } } } }, // Directive
+  { then: 'general' },
+]
+```
+
+Upgrade an AI condition to a code condition by swapping `when` for `if` — the LLM call goes away:
+
+```typescript
+// Before: AI evaluates this — costs tokens
+{ when: 'user is on the enterprise plan', then: 'enterprise_path' }
+
+// After: code evaluates this — free
+{ if: ({ data }) => data.plan === 'enterprise', then: 'enterprise_path' }
+```
+
+---
+
+## Verification
+
+After migrating, confirm no legacy references remain. Run from your repo root:
+
+```bash
+rg -n '\b(Route|RouteOptions|RoutingEngine|END_FLOW|END_ROUTE|ConditionTemplate|EnhancedTool|FlowTransitionConfig|FlowCompletionHandler|pendingTransition|nextStepFlow|transitionTo|endStep|BranchResult|BranchSpec|StepResult|appliedGuidelines|Guideline|Prohibition|skipIf|maxStepsPerBatch|BatchExecutor)\b' \
+  --glob '**/*.ts' \
+  --glob '!node_modules/**' \
+  --glob '!dist/**'
+```
+
+Confirm v1 agent identity fields are gone:
+
+```bash
+rg -n '\b(description|identity|personality)\s*:' \
+  --glob '**/*.ts' \
+  --glob '!node_modules/**' \
+  --glob '!dist/**' \
+  | rg 'createAgent|new Agent\('
+```
+
+Then run the type checker:
+
+```bash
+npx tsc --noEmit
+# or
+bun run typecheck
+```
+
+If both commands return clean (no matches; exit code 0), your migration is complete.
+
+## Cross-References
+
+- [Route → Flow rename](./route-to-flow.md) — the rename pass that ships before v2
+- [CHANGELOG](../../CHANGELOG.md) — full v2.0 release notes