fp-pack 0.9.0 → 0.9.2

package/dist/skills/fp-pack/SKILL.md

@@ -7,1740 +7,433 @@ metadata:
 
 # fp-pack AI Agent Skills
 
-Document Version: 0.9.0
+Document Version: 0.9.2
 
-This document provides guidelines for AI coding assistants when working in projects that use fp-pack. Follow these instructions to write clean, declarative, functional code using fp-pack's utilities.
+## ⚠️ Activation Condition (Read First)
 
-## Project Philosophy
+These guidelines apply **only when `fp-pack` is installed** in the current project.
 
-fp-pack is a TypeScript functional programming library focused on:
+Before following this document:
+- Check `package.json` for `fp-pack` in dependencies/devDependencies
+- Check `node_modules/fp-pack` exists
+- Check existing code imports from `fp-pack` or `fp-pack/stream`
 
-1. **Function Composition**: Use `pipe` and `pipeAsync` as the primary tools for combining operations
-2. **Declarative Code**: Prefer function composition over imperative loops and mutations
-3. **No Monad Pattern**: Traditional FP monads (Option, Either, etc.) are NOT used - they don't compose well with `pipe`
-4. **SideEffect Pattern**: Handle errors and side effects using `SideEffect` with `pipeSideEffect` / `pipeAsyncSideEffect` pipelines (use `pipeSideEffectStrict` / `pipeAsyncSideEffectStrict` for strict unions)
-5. **Lazy Evaluation**: Use `stream/*` functions for efficient iterable processing
+If `fp-pack` is **not** installed, use the project’s existing conventions. Do **not** suggest adding fp-pack unless the user asks.
 
-## Core Composition Functions
+---
+
+## Agent Checklist (TL;DR)
+
+When writing or editing code in an fp-pack project:
+- Prefer `pipe` / `pipeAsync` for composition; keep helpers **unary** in the pipeline.
+- Prefer **data-last** utilities (curried) so they compose: `map(fn)`, `prop('k')`, `assoc('k', v)`, etc.
+- Use `stream/*` for lazy/large/iterable processing; use array/object utils for small/eager data.
+- Use SideEffect-aware pipes only when you need **early exit**: `pipeSideEffect*` / `pipeAsyncSideEffect*`.
+- **Never call `runPipeResult` inside a pipeline**; call it at the boundary (event handler, service wrapper, etc.).
+- If TypeScript inference gets stuck with a data-last generic, use a small wrapper, an explicit type hint, or `pipeHint`.
+
+## Philosophy (What fp-pack Optimizes For)
+
+fp-pack is a TypeScript FP utility library focused on:
+1) Function composition via `pipe` / `pipeAsync`
+2) Declarative, immutable transforms (avoid loops and mutations)
+3) Practical TypeScript inference (data-last + currying for most helpers)
+4) Optional early-exit handling via `SideEffect` (only when needed)
+5) Lazy iterable processing via `fp-pack/stream` for performance-sensitive flows
+
+---
+
+## Core Composition
 
-### pipe - Synchronous Function Composition
+### `pipe` (sync)
 
-**Always prefer `pipe` for synchronous operations** instead of manual imperative code.
+Use `pipe` to build a unary function:
 
-```typescript
-import { pipe, map, filter, take } from 'fp-pack';
+```ts
+import { pipe, filter, map, take } from 'fp-pack';
 
-// GOOD: Declarative pipe composition
 const processUsers = pipe(
-  filter((user: User) => user.age >= 18),
-  map(user => user.name.toUpperCase()),
+  filter((u: User) => u.active),
+  map((u) => u.name),
   take(10)
 );
-
-// BAD: Imperative approach
-const processUsers = (users: User[]) => {
-  const result = [];
-  for (const user of users) {
-    if (user.age >= 18) {
-      result.push(user.name.toUpperCase());
-      if (result.length >= 10) break;
-    }
-  }
-  return result;
-};
 ```
 
-> For SideEffect-based early exits, use `pipeSideEffect` (or `pipeSideEffectStrict` when you want strict unions).
+### `pipeAsync` (async)
 
-### pipeAsync - Asynchronous Function Composition
+Use `pipeAsync` when any step is async:
 
-**Use `pipeAsync` for any async operations** including API calls, database queries, or async transformations.
-
-```typescript
+```ts
 import { pipeAsync } from 'fp-pack';
 
-// GOOD: Async pipe composition
-const fetchUserData = pipeAsync(
-  async (userId: string) => fetch(`/api/users/${userId}`),
-  async (response) => response.json(),
+const fetchUser = pipeAsync(
+  async (id: string) => fetch(`/api/users/${id}`),
+  async (res) => res.json(),
   (data) => data.user
 );
-
-// BAD: Manual async handling
-const fetchUserData = async (userId: string) => {
-  const response = await fetch(`/api/users/${userId}`);
-  const data = await response.json();
-  return data.user;
-};
 ```
 
-> For SideEffect-aware async pipelines, use `pipeAsyncSideEffect` (or `pipeAsyncSideEffectStrict` for strict unions).
+---
+
+## Currying & Data-Last Conventions
+
+Most multi-arg helpers are **data-last** and **curried**, so they work naturally in `pipe`:
+
+- Good: `map(fn)`, `filter(pred)`, `replace(from, to)`, `assoc('k', v)`, `path(['a','b'])`
+- Avoid in pipelines: helpers that require you to pass the data first, or that return a function whose type depends on the final data arg (see next section)
+
+Single-arg helpers are already unary; “currying them” adds no value—just use them directly.
+
+---
+
+## TypeScript: Data-last Generic Inference Caveats (Important)
 
-## Data-last Generic Inference Caveats
+Some data-last helpers return a **generic function** whose type is only determined by the final data argument.
+Inside `pipe`/`pipeAsync`, TypeScript sometimes can’t infer that type, so you may need a tiny hint.
 
-Some data-last helpers return a **generic function** whose type is determined only by the final data argument. Inside
-`pipe`/`pipeAsync`, TypeScript cannot infer that type, so you may need `pipeHint`, a small type hint, or a data-first wrapper.
+### Recommended fixes (choose 1)
 
-```typescript
+```ts
 import { pipe, pipeHint, zip, some } from 'fp-pack';
 
-// Option 1: data-first wrapper
+// 1) data-first wrapper (most explicit)
 const withWrapper = pipe(
   (values: number[]) => zip([1, 2, 3], values),
   some(([a, b]) => a > b)
 );
 
-// Option 2: explicit type annotation
+// 2) explicit function type hint (short, but uses an assertion)
 const withHint = pipe(
   zip([1, 2, 3]) as (values: number[]) => Array<[number, number]>,
   some(([a, b]) => a > b)
 );
 
-// Option 3: pipeHint helper
+// 3) pipeHint helper (keeps pipeline style)
 const withPipeHint = pipe(
   pipeHint<number[], Array<[number, number]>>(zip([1, 2, 3])),
   some(([a, b]) => a > b)
 );
 ```
 
-**Utilities that may need a type hint in data-last pipelines:**
-- **Array**: `chunk`, `drop`, `take`, `zip`
-- **Object**: `assoc`, `assocPath`, `dissocPath`, `evolve`, `mapValues`, `merge`, `mergeDeep`, `omit`, `path`, `pathOr`, `pick`, `prop`, `propOr`, `propStrict`
-- **Async**: `timeout`
-- **Stream**: `chunk`, `drop`, `take`, `zip`
+### Utilities that may need a hint in data-last pipelines
 
-## SideEffect Pattern - For Special Cases Only
+- Array: `chunk`, `drop`, `take`, `zip`
+- Object: `assoc`, `assocPath`, `dissocPath`, `evolve`, `mapValues`, `merge`, `mergeDeep`, `omit`, `path`, `pathOr`, `pick`, `prop`, `propOr`, `propStrict`
+- Async: `timeout`
+- Stream: `chunk`, `drop`, `take`, `zip`
 
-**Most cases: Use `pipe` / `pipeAsync` - they're simpler and sufficient for 99% of use cases.**
+When in doubt: check the `.d.ts` signature (see “Quick Signature Lookup”).
 
-`pipe` and `pipeAsync` are for **pure** functions and don't handle `SideEffect`. **Only use `pipeSideEffect`/`pipeAsyncSideEffect` when you specifically need**:
-- Early termination based on validation
-- Error handling with side effects (logging, toasts, etc.)
-- Optional chaining patterns
-
-For regular error handling, standard try-catch or error propagation is perfectly fine.
-If you want precise SideEffect unions across branches, use `pipeSideEffectStrict` / `pipeAsyncSideEffectStrict`.
+---
 
-```typescript
-// MOST CASES: Just use pipe with regular error handling
-import { pipe, map, filter } from 'fp-pack';
+## SideEffect Pattern (Use Only When Needed)
 
-const processData = pipe(
-  validateInput,
-  transformData,
-  saveData
-);
+Most code should use `pipe` / `pipeAsync`. Use SideEffect-aware pipes only when you truly need **early termination**:
+- validation pipelines that should stop early
+- recoverable errors you want to model as data
+- branching flows where you want to short-circuit
 
-try {
-  const result = processData(input);
-} catch (error) {
-  console.error('Processing failed:', error);
-}
+### SideEffect-aware pipes
+- `pipeSideEffect` / `pipeAsyncSideEffect`: convenient, but may widen effects to `any`
+- `pipeSideEffectStrict` / `pipeAsyncSideEffectStrict`: preserves strict union effects (recommended when you care about types)
 
-// SPECIAL CASES: Use pipeSideEffect when you need early termination with side effects
-import { pipeSideEffect, SideEffect, runPipeResult } from 'fp-pack';
-
-const processDataPipeline = pipeSideEffect(
-  validateInput,
-  (data) => {
-    if (!data.isValid) {
-      return SideEffect.of(() => {
-        showToast('Invalid data');  // Side effect
-        logError('validation_failed');  // Side effect
-        return null;
-      });
-    }
-    return data;
-  },
-  transformData
-);
+### Rules
+- Do **not** call `runPipeResult` or `matchSideEffect` **inside** the pipeline.
+- Prefer `isSideEffect` for precise runtime narrowing in both branches.
+- Use `runPipeResult` only at the boundary, and provide generics if inference is widened.
 
-// runPipeResult must be called OUTSIDE the pipeline
-const finalValue = runPipeResult(processDataPipeline(input));
-```
+### Key functions
+- `SideEffect.of(effectFn, label?)`
+- `isSideEffect(value)` (type guard)
+- `runPipeResult(result)` (execute effect or return value; **outside** pipelines)
+- `matchSideEffect(result, { value, effect })`
 
-**Key SideEffect functions:**
-- `SideEffect.of(fn, label?)` - Create a side effect container
-- `isSideEffect(value)` - Type guard for **runtime checking** whether a value is a SideEffect
-- `runPipeResult<T, R>(result)` - Execute SideEffect or return value (call **OUTSIDE** pipelines). If the input is narrowed to `SideEffect<R>` (e.g. after `isSideEffect`), it returns `R`. If the input is widened to `SideEffect<any>` or `any`, the result becomes `any` unless you provide generics.
-- `matchSideEffect(result, { value, effect })` - Pattern match on result
+### `runPipeResult` type behavior (important)
 
-**Type-safe result handling:**
+- If the input is **narrowed** to `SideEffect<R>` (e.g. inside `if (isSideEffect(x))`), `runPipeResult(x)` returns `R`.
+- If the input is **widened** to `SideEffect<any>` or `any` (common with non-strict pipelines), `runPipeResult(x)` becomes `any` unless you provide generics.
 
-```typescript
-import { pipeSideEffect, pipeSideEffectStrict, SideEffect, isSideEffect, runPipeResult } from 'fp-pack';
+```ts
+import {
+  pipeSideEffect,
+  pipeSideEffectStrict,
+  SideEffect,
+  isSideEffect,
+  runPipeResult,
+} from 'fp-pack';
 
-const processNumbers = pipeSideEffectStrict(
-  (nums: number[]) => nums.filter(n => n % 2 === 1),
-  (odds) => odds.length > 0
-    ? odds
-    : SideEffect.of(() => 'No odd numbers'),
-  (odds) => odds.map(n => n * 2)
+const nonStrict = pipeSideEffect(
+  (n: number) => (n > 0 ? n : SideEffect.of(() => 'NEG' as const))
 );
+const widened: number | SideEffect<any> = nonStrict(-1);
+const unsafe = runPipeResult(widened); // any
 
-const result = processNumbers([1, 2, 3, 4, 5]);
+const strict = pipeSideEffectStrict(
+  (n: number) => (n > 0 ? n : SideEffect.of(() => 'NEG' as const))
+);
+const precise = strict(-1); // number | SideEffect<'NEG'>
 
-// ✅ CORRECT: Use isSideEffect for runtime checking
-if (!isSideEffect(result)) {
-  // TypeScript knows: result is number[]
-  const sum: number = result.reduce((a, b) => a + b, 0);
-} else {
-  // TypeScript knows: result is SideEffect<'No odd numbers'>
-  const error = runPipeResult(result);  // 'No odd numbers'
+if (isSideEffect(precise)) {
+  const err = runPipeResult(precise); // 'NEG'
 }
-
-// ⚠️ Non-strict pipeSideEffect widens SideEffect to any
-const widened: number[] | SideEffect<any> = pipeSideEffect(
-  (nums: number[]) => nums,
-  (nums) => nums.length > 0 ? nums : SideEffect.of(() => 'EMPTY')
-)([]);
-const unsafeValue = runPipeResult(widened);  // any
-
-// ✅ CORRECT: Provide generics to recover a safe union
-const safeValue = runPipeResult<number[], string>(result);  // result: number[] | string (union type - safe but not narrowed)
 ```
 
-**⚠️ CRITICAL: runPipeResult Type Safety**
-
-`runPipeResult<T, R=any>` has a default type parameter `R=any`. This means:
-
-- ✅ **Precise input types**: `T | SideEffect<'E'>` preserves `T | 'E'` without extra annotations.
-- ⚠️ **Widened inputs**: `T | SideEffect<any>` (or `any`) collapses to `any`.
-- ✅ **With generics**: `runPipeResult<SuccessType, ErrorType>(result)` restores a safe union when inference is lost.
-- ✅ **After narrowing**: If the input is `SideEffect<'E'>`, `runPipeResult` returns `'E'`.
-- ✅ **With isSideEffect**: Prefer for runtime narrowing when you need branch-specific types.
+---
 
-Provide generics when inference is lost; prefer `isSideEffect` for precise narrowing.
+## Stream Functions (`fp-pack/stream`) — Lazy Iterables
 
-## Stream Functions - Lazy Iterable Processing
+Use stream utilities when:
+- data is large or unbounded (`range`, streams, generators)
+- you want lazy evaluation (avoid allocating intermediate arrays)
+- you want to support `Iterable` and `AsyncIterable`
 
-**Use `stream/*` functions for lazy, memory-efficient data processing** instead of array methods.
+Stream utilities are in `fp-pack/stream` and are designed to be pipe-friendly.
 
-```typescript
+```ts
 import { pipe } from 'fp-pack';
-import { map, filter, take, toArray, range } from 'fp-pack/stream';
+import { range, filter, map, take, toArray } from 'fp-pack/stream';
 
-// GOOD: Lazy stream processing
-const processLargeDataset = pipe(
+const first100SquaresOfEvens = pipe(
   filter((n: number) => n % 2 === 0),
-  map(n => n * n),
+  map((n) => n * n),
   take(100),
   toArray
 );
 
-// Processes only what's needed - memory efficient
-const result = processLargeDataset(range(1, 1000000));
-
-// BAD: Eager array processing
-const result = Array.from({ length: 1000000 }, (_, i) => i + 1)
-  .filter(n => n % 2 === 0)
-  .map(n => n * n)
-  .slice(0, 100); // Processed entire dataset!
+const result = first100SquaresOfEvens(range(Infinity));
 ```
 
-**Stream functions support both sync and async iterables:**
-- Sync: `Iterable<T>` → `IterableIterator<R>`
-- Async: `AsyncIterable<T>` → `AsyncIterableIterator<R>`
+---
 
-## Available Functions by Category
+## Available Functions (Practical Index)
 
-### Composition
-- `pipe` - Left-to-right function composition (sync)
-- `pipeSideEffect` - Left-to-right composition with SideEffect short-circuiting
-- `pipeSideEffectStrict` - SideEffect composition with strict effect unions
-- `compose` - Right-to-left function composition
-- `curry` - Curry a function
-- `partial` - Partial application
-- `flip` - Flip function argument order
-- `complement` - Logical negation
-- `identity` - Return input unchanged
-- `constant` - Always return the same value
-- `from` - Ignore input and return a fixed value
-- `tap` - Execute side effect and return original value
-- `tap0` - Execute side effect without input
-- `once` - Execute function only once
-- `memoize` - Cache function results
-- `SideEffect` - Side effect container
-- `isSideEffect` - Type guard for SideEffect
-- `matchSideEffect` - Pattern match on value/SideEffect
-- `runPipeResult` - Execute SideEffect or return value
+This is not a complete API reference; it’s the “what to reach for” index when composing pipelines.
 
-### Async
-- `pipeAsync` - Async function composition
-- `pipeAsyncSideEffect` - Async composition with SideEffect short-circuiting
-- `pipeAsyncSideEffectStrict` - Async SideEffect composition with strict effect unions
-- `delay` - Delay execution
-- `timeout` - Add timeout to promise
-- `retry` - Retry failed operations
-- `debounce` - Debounce function calls
-- `debounceLeading` - Debounce with leading edge
-- `debounceLeadingTrailing` - Debounce with both edges
-- `throttle` - Throttle function calls
+### Composition
+- `pipe`, `pipeAsync`
+- SideEffect-aware: `pipeSideEffect`, `pipeSideEffectStrict`, `pipeAsyncSideEffect`, `pipeAsyncSideEffectStrict`
+- Utilities: `from`, `tap`, `tap0`, `once`, `memoize`, `identity`, `constant`, `flip`, `partial`, `curry`, `compose`, `complement`
+- SideEffect helpers: `SideEffect`, `isSideEffect`, `matchSideEffect`, `runPipeResult`
 
 ### Array
-- `map`, `filter`, `reduce`, `flatMap`
-- `find`, `some`, `every`
-- `take`, `drop`, `takeWhile`, `dropWhile`
-- `chunk`, `zip`, `zipWith`, `unzip`, `zipIndex`
-- `uniq`, `uniqBy`, `sort`, `sortBy`, `groupBy`
-- `concat`, `append`, `prepend`, `flatten`, `flattenDeep`
-- `head`, `tail`, `last`, `init`
-- `range`, `partition`, `scan`
+- Core transforms: `map`, `filter`, `flatMap`, `reduce`, `scan`
+- Queries: `find`, `some`, `every`
+- Slicing: `take`, `drop`, `takeWhile`, `dropWhile`, `chunk`
+- Ordering/grouping: `sort`, `sortBy`, `groupBy`, `uniqBy`
+- Pairing: `zip`, `zipWith`
+- Combining: `concat`, `append`, `prepend`, `flatten`
 
 ### Object
-- `prop`, `propOr`, `path`, `pathOr`
-- `pick`, `omit`
-- `assoc`, `assocPath`, `dissoc`, `dissocPath`
-- `merge`, `mergeDeep`, `mergeAll`
-- `keys`, `values`, `entries`
-- `mapValues`, `evolve`
-- `has`, `hasPath`
+- Access: `prop`, `propOr`, `propStrict`, `path`, `pathOr`
+- Pick/drop: `pick`, `omit`
+- Updates: `assoc`, `assocPath`, `dissocPath`
+- Merge: `merge`, `mergeDeep`
+- Transforms: `mapValues`, `evolve`
+- Predicates: `has`
 
 ### Control Flow
-- `ifElse` - Conditional branching
-- `when`, `unless` - Conditional execution
-- `cond` - Multi-branch conditional
-- `tryCatch` - Safe function execution
-- `guard` - Validation guard
-
-### Stream (Lazy Iterables)
-- `append`, `concat`, `prepend`
-- `map`, `filter`, `flatMap`, `flatten`, `flattenDeep`
-- `take`, `takeWhile`, `drop`, `dropWhile`, `chunk`
-- `zip`, `zipWith`, `find`, `some`, `every`
-- `reduce`, `scan`
-- `range`
-- `toArray` - Materialize stream to array
-- `toAsync` - Convert to async iterable
-
-### Math
-- `add`, `sub`, `mul`, `div`
-- `sum`, `mean`, `min`, `max`
-- `round`, `floor`, `ceil`, `randomInt`
-
-### String
-- `trim`, `split`, `join`, `replace`
-- `toUpper`, `toLower`
-- `startsWith`, `endsWith`, `match`
-
-### Equality
-- `equals`, `includes`
-- `isNil`, `isEmpty`, `isType`
-- `gt`, `gte`, `lt`, `lte`
-- `clamp`
-
-### Nullable
-- `maybe`, `mapMaybe`, `getOrElse`, `fold`, `result`
-
-### Debug
-- `assert`, `invariant`, `log`
+- `ifElse`, `when`, `unless`, `cond`, `guard`, `tryCatch`
 
-## Coding Guidelines for AI Agents
-
-### 0. Detect Project Language (JS vs TS)
-
-Before writing code, check whether the project is JavaScript or TypeScript:
-
-- **TypeScript projects**: use explicit types, leverage generics, and keep type-safe signatures in examples.
-- **JavaScript projects**: avoid TypeScript-only syntax and prefer JSDoc only when it adds clarity or is already used.
-
-### 0.1 Quick Signature Lookup
-
-If a function signature or argument order is unclear, check the local declaration or source files:
-
-- Main exports: `dist/index.d.ts`
-- Stream exports: `dist/stream/index.d.ts`
-- Main utilities (fallback): `src/implement/**`
-- Stream utilities (fallback): `src/stream/**`
-- Installed package:
-  - `node_modules/fp-pack/dist/index.d.ts`
-  - `node_modules/fp-pack/dist/stream/index.d.ts`
-  - `node_modules/fp-pack/src/implement/**`
-  - `node_modules/fp-pack/src/stream/**`
-
-### 1. Always Prefer pipe/pipeAsync
-
-```typescript
-// GOOD
-const result = pipe(
-  trim,
-  split(','),
-  map(toNumber),
-  filter(isPositive)
-)(input);
-
-// BAD
-const trimmed = trim(input);
-const parts = split(',')(trimmed);
-const numbers = map(toNumber)(parts);
-const result = filter(isPositive)(numbers);
-```
-
-### 2. Use Curried Functions (Where Available)
-
-Most multi-arg functions are curried. Single-arg utilities are already unary, so currying adds no benefit; use them directly (e.g. `uniq`, `flatten`, `flattenDeep`, `head`, `tail`, `last`, `init`, `range`, `sum`, `mean`, `min`, `max`, `round`, `floor`, `ceil`, `trim`, `toLower`, `toUpper`, `isNil`, `isEmpty`, `isType`). These are already pipe-friendly without a curried variant.
+### Async
+- `retry`, `timeout`, `delay`
+- Function-returning helpers: `debounce*`, `throttle`
 
-```typescript
-import { pipe, map, filter } from 'fp-pack';
+### Stream (Lazy Iterables)
+- Building: `range`
+- Transforms: `map`, `filter`, `flatMap`, `flatten`
+- Slicing: `take`, `drop`, `takeWhile`, `dropWhile`, `chunk`
+- Queries/reductions: `find`, `some`, `every`, `reduce`, `scan`
+- Pairing/combining: `zip`, `zipWith`, `concat`, `append`, `prepend`
+- Utilities: `toArray`, `toAsync`
+
+### Math / String / Equality / Debug
+- Math: `add`, `sub`, `mul`, `div`, `randomInt`, `clamp`
+- String: `split`, `join`, `replace`, `match`, `trim`
+- Equality: `equals`, `isNil`
+- Debug: `assert`, `invariant`
 
-// GOOD: Curried usage in pipe
-const processUsers = pipe(
-  filter(user => user.active),
-  map(user => user.name)
-);
+---
 
-// GOOD: Partial application
-const filterActive = filter((user: User) => user.active);
-const getNames = map((user: User) => user.name);
-const processUsers = pipe(filterActive, getNames);
-```
+## Micro-Patterns (Keep It Short)
 
-### 2.1 Custom Utility Authoring (Curry Typing)
+### Boundary handling (UI/event/service)
 
-When you add your own helpers for `pipe`, follow these rules:
+Do the work in a pipeline, unwrap at the boundary:
 
-- Keep **data-last** argument order.
-- **Curry multi-arg functions** so they compose well.
-- **Fixed signatures** can use `curry(fn)` directly.
-- **Generic or overloaded signatures** should use an explicit type alias + cast to preserve inference.
+```ts
+import { pipeSideEffectStrict, SideEffect, isSideEffect, runPipeResult } from 'fp-pack';
 
-```typescript
-// Fixed signature: curry is enough
-function split(separator: string, str: string): string[] {
-  return str.split(separator);
-}
-export default curry(split);
+const validate = (n: number) => (n > 0 ? n : SideEffect.of(() => 'NEG' as const));
+const pipeline = pipeSideEffectStrict(validate, (n) => n + 1);
 
-// Generic signature: add a type alias for the curried form
-type Chunk = {
-  (size: number): <T>(arr: T[]) => T[][];
-  <T>(size: number, arr: T[]): T[][];
+export const handler = (n: number) => {
+  const result = pipeline(n);
+  if (isSideEffect(result)) return runPipeResult(result); // 'NEG'
+  return result; // number
 };
-
-function chunk<T>(size: number, arr: T[]): T[][] {
-  // ...
-  return [];
-}
-
-const curriedChunk = curry(chunk) as Chunk;
-export default curriedChunk;
 ```
 
-### 3. Choose pipe vs pipeSideEffect
-
-**Default choice: Start with `pipe` / `pipeAsync`**
-
-Most data transformations are pure and don't need SideEffect handling. Use `pipe` for sync operations and `pipeAsync` for async operations. **Only switch to SideEffect-aware pipes when you actually need** early termination or error handling with side effects.
+### Data-first style with `from()`
 
-- **`pipe`** - Synchronous, **pure** transformations (99% of cases)
-- **`pipeAsync`** - Async, **pure** transformations (99% of cases)
-- **`pipeSideEffect`** - **Only when you need** SideEffect short-circuiting (sync)
-- **`pipeAsyncSideEffect`** - **Only when you need** SideEffect short-circuiting (async)
-- **`pipeSideEffectStrict`** - Sync SideEffect pipelines with strict effect unions
-- **`pipeAsyncSideEffectStrict`** - Async SideEffect pipelines with strict effect unions
+Some agents/users prefer “data-first” pipelines. You can keep `pipe` data-last and still write data-first flows by injecting the initial value with `from()`:
 
-**Important:** `pipe` and `pipeAsync` are for **pure** functions only—they don't handle `SideEffect`. If your pipeline can return `SideEffect`, use `pipeSideEffect` or `pipeAsyncSideEffect` instead. Choose the strict variants when you need precise unions for SideEffect results.
+```ts
+import { pipe, from, map, filter, take } from 'fp-pack';
 
-```typescript
-// Sync: use pipe
-const processNumbers = pipe(
-  map((n: number) => n * 2),
-  filter(n => n > 10)
-);
+// data-first feeling: start with data, then compose transforms
+const result = pipe(
+  from([1, 2, 3, 4, 5]),
+  filter((n: number) => n % 2 === 0),
+  map((n) => n * 10),
+  take(2)
+)(); // [20, 40]
 
-// Async: use pipeAsync
-const processUsers = pipeAsync(
-  async (ids: string[]) => db.users.findMany(ids),
-  map(user => user.profile),
-  filter(profile => profile.verified)
+// same idea when you already have an input (normal usage)
+const process = pipe(
+  filter((n: number) => n % 2 === 0),
+  map((n) => n * 10)
 );
+const result2 = process([1, 2, 3, 4, 5]); // [20, 40]
 ```
 
-### 3.1. SideEffect Composition Rule
-
-**🔄 Critical Rule: SideEffect Contagion**
-
-Once you use `pipeSideEffect` or `pipeAsyncSideEffect`, the result is **always `T | SideEffect`** (or `Promise<T | SideEffect>` for async). The same rule applies to strict variants.
-
-If you want to continue composing this result, you **MUST** keep using SideEffect-aware pipes. You **CANNOT** switch back to `pipe` or `pipeAsync` because they don't handle `SideEffect`.
+### `ifElse` uses functions (not values)
 
-```typescript
-import { pipe, pipeSideEffect, SideEffect } from 'fp-pack';
+```ts
+import { ifElse, from } from 'fp-pack';
 
-const validateUserPipeline = pipeSideEffect(
-  findUser,
-  validateAge
-);
-// Result type: User | SideEffect
-
-// ❌ WRONG - pipe cannot handle SideEffect
-const wrongPipeline = pipe(
-  validateUserPipeline,  // Returns User | SideEffect
-  (user) => user.email   // Type error! SideEffect has no 'email' property
-);
-
-// ✅ CORRECT - Keep using pipeSideEffect
-const correctPipeline = pipeSideEffect(
-  validateUserPipeline,  // User | SideEffect - handled correctly
-  (user) => user.email,  // Automatically skipped if SideEffect
-  sendEmail
-);
-
-// The same rule applies to async pipes
-const asyncPipeline = pipeAsyncSideEffect(
-  fetchUser,
-  validateUser
-);
-// Result type: Promise<User | SideEffect>
-
-// You must continue with pipeAsyncSideEffect, not pipeAsync
-const extendedAsyncPipeline = pipeAsyncSideEffect(
-  asyncPipeline,
-  processUser,
-  saveToDatabase
+const label = ifElse(
+  (n: number) => n >= 60,
+  from('pass'),
+  from('fail')
 );
 ```
 
-### 4. Use stream/* for Large Datasets
+### Stream pipeline to array
 
-```typescript
+```ts
 import { pipe } from 'fp-pack';
-import { filter, map, take, toArray, range } from 'fp-pack/stream';
+import { map, filter, toArray } from 'fp-pack/stream';
 
-// GOOD: Lazy processing
-const getFirst100Even = pipe(
-  filter((n: number) => n % 2 === 0),
-  take(100),
+const toIds = pipe(
+  filter((u: User) => u.active),
+  map((u) => u.id),
   toArray
 );
-
-// Stops after finding 100 items (only processes 100, not 1 million)
-const result = getFirst100Even(range(1, 1000000));
 ```
 
-### 5. Handle Errors with SideEffect
-
-```typescript
-import { pipeSideEffect, SideEffect, runPipeResult } from 'fp-pack';
-
-const safeDividePipeline = pipeSideEffect(
-  (input: { a: number; b: number }) => {
-    if (input.b === 0) {
-      return SideEffect.of(() => {
-        throw new Error('Division by zero');
-      }, 'DIVISION_ERROR');
-    }
-    return input;
-  },
-  ({ a, b }) => a / b
-);
+### Data-last inference workaround (when needed)
 
-// runPipeResult must be called OUTSIDE the pipeline
-const result = runPipeResult(safeDividePipeline({ a: 10, b: 2 })); // 5
-```
+Use a wrapper or `pipeHint` when a data-last generic won’t infer inside `pipe`:
 
-### 6. Use Control Flow Functions
+```ts
+import { pipe, pipeHint, zip } from 'fp-pack';
 
-```typescript
-import { pipe, ifElse, when, cond } from 'fp-pack';
+// wrapper
+const zipUsers = pipe((xs: User[]) => zip(['a', 'b', 'c'], xs));
 
-// GOOD: Declarative conditionals
-const processAge = pipe(
-  ifElse(
-    (age: number) => age >= 18,
-    age => ({ age, status: 'adult' }),
-    age => ({ age, status: 'minor' })
-  )
+// pipeHint
+const zipUsers2 = pipe(
+  pipeHint<User[], Array<[string, User]>>(zip(['a', 'b', 'c']))
 );
-
-// GOOD: Multi-branch with cond
-const gradeToLetter = cond([
-  [(n: number) => n >= 90, () => 'A'],
-  [(n: number) => n >= 80, () => 'B'],
-  [(n: number) => n >= 70, () => 'C'],
-  [() => true, () => 'F']
-]);
 ```
 
-### 6.1 Type-Safety Tips (prop/ifElse/cond)
-
-- `prop` returns `T[K] | undefined`. Use `propOr` (or guard) before array operations.
-- `ifElse` expects **functions** for both branches. If you already have a value, wrap it: `() => value` or use `from(value)` for cleaner constant branches.
-- Use `from(value)` when you need a unary function that ignores input (handy for `ifElse`/`cond` branches and data-first patterns). Pipelines that start with `from(...)` can be called without an initial input value.
-- `cond` returns `R | undefined`. Add a default branch and coalesce when you need a strict result.
-- In `pipeSideEffect`, keep step return types aligned to avoid wide unions.
-
-```typescript
-import { pipe, propOr, append, assoc, ifElse, cond, from, filter, map } from 'fp-pack';
-
-// propOr keeps the type strict for array ops
-const addTodo = (text: string, state: AppState) =>
-  pipe(
-    propOr([], 'todos'),
-    append(createTodo(text)),
-    (todos) => assoc('todos', todos, state)
-  )(state);
-
-// ifElse expects functions, not values
-const toggleTodo = (id: string) => ifElse(
-  (todo: Todo) => todo.id === id,
-  assoc('completed', true),
-  (todo) => todo
-);
-
-// from is useful for constant branches - cleaner than () => value
-const statusLabel = ifElse(
-  (score: number) => score >= 60,
-  from('pass'),    // Constant value
-  from('fail')
-);
-
-// Data-first pattern with from: inject data into pipeline
-const processData = pipe(
-  from([1, 2, 3, 4, 5]),
-  filter((n: number) => n % 2 === 0),
-  map(n => n * 2)
-);
-const result = processData(); // [4, 8]
-
-// cond still returns R | undefined, so coalesce if needed
-const grade = (score: number) =>
-  cond([
-    [(n: number) => n >= 90, () => 'A'],
-    [() => true, () => 'F']
-  ])(score) ?? 'F';
-```
+### Async pipelines (retry/timeout)
 
-### 7. Object Transformations
+Keep async steps inside `pipeAsync` and push configuration (like `timeout`) via currying:
 
-```typescript
-import { pipe, pick, mapValues, merge } from 'fp-pack';
+```ts
+import { pipeAsync, retry, timeout } from 'fp-pack';
 
-// GOOD: Declarative object operations
-const processUser = pipe(
-  pick(['name', 'email', 'age']),
-  mapValues((value) => typeof value === 'string' ? value.trim() : value),
-  merge({ verified: false })
+const fetchJson = pipeAsync(
+  (url: string) => fetch(url),
+  (res) => res.json()
 );
-```
-
-## Anti-Patterns to Avoid
-
-### ❌ Don't use imperative loops
 
-```typescript
-// BAD
-const result = [];
-for (const item of items) {
-  if (item.active) {
-    result.push(item.name);
-  }
-}
-
-// GOOD
-const result = pipe(
-  filter((item: Item) => item.active),
-  map(item => item.name)
-)(items);
+const fetchJsonWithGuards = pipeAsync(
+  fetchJson,
+  timeout(5_000),
+  retry(3)
+);
 ```
 
-### ❌ Don't chain array methods
+### Object updates (avoid mutation)
 
-```typescript
-// BAD
-const result = users
-  .filter(u => u.active)
-  .map(u => u.name)
-  .slice(0, 10);
+```ts
+import { pipe, assocPath, merge } from 'fp-pack';
 
-// GOOD
-const result = pipe(
-  filter((u: User) => u.active),
-  map(u => u.name),
-  take(10)
-)(users);
-```
-
-### ❌ Don't use traditional monads (Option, Either, Maybe)
-
-```typescript
-// BAD - Don't implement this pattern
-const maybeUser = Option.of(user)
-  .map(u => u.profile)
-  .flatMap(p => p.email);
-
-// GOOD - Use SideEffect with pipeSideEffect
-const getUserEmail = pipeSideEffect(
-  (user: User) => {
-    if (!user.profile) {
-      return SideEffect.of(() => null, 'NO_PROFILE');
-    }
-    return user.profile;
-  },
-  (profile) => {
-    if (!profile.email) {
-      return SideEffect.of(() => null, 'NO_EMAIL');
-    }
-    return profile.email;
-  }
+const updateAccount = pipe(
+  assocPath(['profile', 'role'], 'member'),
+  merge({ updatedAt: Date.now() })
 );
 ```
 
-### ❌ Don't mutate data
+---
 
-```typescript
-// BAD
-const updateUser = (user: User) => {
-  user.lastLogin = new Date();
-  return user;
-};
+## Decision Guide (What Should the Agent Pick?)
 
-// GOOD
-const updateUser = (user: User) => ({
-  ...user,
-  lastLogin: new Date()
-});
+- Is everything sync and pure? → `pipe`
+- Any step async? → `pipeAsync`
+- Need early-exit + typed effect unions? → `pipeSideEffectStrict` / `pipeAsyncSideEffectStrict`
+- Need early-exit but type precision doesn’t matter? → `pipeSideEffect` / `pipeAsyncSideEffect`
+- Handling result at boundary?
+  - Need exact branch types? → `isSideEffect` + separate branches
+  - Just want to “unwrap” and don’t care about precision? → `runPipeResult` (provide generics if widened)
+- Large/unbounded/iterable data? → `fp-pack/stream`
 
-// EVEN BETTER with fp-pack
-import { assoc } from 'fp-pack';
-const updateUser = assoc('lastLogin', new Date());
-```
+---
 
 ## Quick Reference
 
 ### Import Paths
-- Main functions: `import { pipe, map, filter } from 'fp-pack'`
-- Async: `import { pipeAsync, delay, retry } from 'fp-pack'`
-- SideEffect: `import { pipeSideEffect, pipeSideEffectStrict, pipeAsyncSideEffect, pipeAsyncSideEffectStrict, SideEffect } from 'fp-pack'`
+
+- Main: `import { pipe, map, filter } from 'fp-pack'`
+- SideEffect pipes: `import { pipeSideEffect, pipeSideEffectStrict, SideEffect } from 'fp-pack'`
+- Async: `import { pipeAsync, retry, timeout } from 'fp-pack'`
 - Stream: `import { map, filter, toArray } from 'fp-pack/stream'`
 
 ### When to Use What
-- **Pure sync transformations**: `pipe` + array/object functions
-- **Pure async operations**: `pipeAsync`
-- **Error handling with SideEffect**: `pipeSideEffect` (sync) / `pipeAsyncSideEffect` (async)
-- **Strict SideEffect unions**: `pipeSideEffectStrict` (sync) / `pipeAsyncSideEffectStrict` (async)
-- **Type-safe result handling**: `isSideEffect` for precise type narrowing (prefer this when you need branch-specific types)
-- **Execute SideEffect**: `runPipeResult` (call OUTSIDE pipelines). If the input is narrowed to `SideEffect<R>`, it returns `R`. If the input is widened to `SideEffect<any>`/`any`, the result becomes `any`; provide generics to recover
-- **Large datasets**: `stream/*` functions
-- **Conditionals**: `ifElse`, `when`, `unless`, `cond`
-- **Object access**: `prop`, `propStrict`, `path`, `pick`, `omit`
-- **Object updates**: `assoc`, `merge`, `evolve`
-
-## UI Framework Integration Patterns
-
-fp-pack works seamlessly with UI frameworks. Here are common patterns organized by **use case**, not framework.
-
-### Pattern 1: Handling User Input
-
-**When**: Form inputs, button clicks, drag & drop, any user interaction
-**Where to use**: Event handlers (onChange, @input, on:click, etc.)
-
-```typescript
-import { pipe, pipeAsyncSideEffect, trim, prop, assoc, tap, SideEffect, runPipeResult } from 'fp-pack';
-
-// GOOD: Process form input declaratively
-const handleNameChange = pipe(
-  prop('currentTarget'),           // Safer than target in most UI libs
-  (el) => (el as HTMLInputElement).value,
-  trim,
-  tap((value) => {
-    // Prefer updater form to avoid stale state in React-like frameworks
-    setFormState(prev => assoc('name', value, prev));
-  })
-);
 
-// Use in any framework:
-// React: <input onChange={handleNameChange} />
-// Vue: <input @input="handleNameChange" />
-// Svelte: <input on:input={handleNameChange} />
-
-// GOOD: Complex form validation
-const validateFieldsOrStop = (data: any) => {
-  const errors = validateFields(data);
-  if (!errors) return data;
-  return SideEffect.of(() => {
-    setErrors(errors);
-    return null;
-  }, 'VALIDATION_ERROR');
-};
+- Pure sync transforms: `pipe`
+- Pure async transforms: `pipeAsync`
+- Early-exit pipelines: `pipeSideEffect*` / `pipeAsyncSideEffect*`
+- Strict effect unions: prefer `pipeSideEffectStrict` / `pipeAsyncSideEffectStrict`
+- Runtime branching: prefer `isSideEffect`
+- Boundary unwrapping: `runPipeResult` (provide generics if inference is widened)
+- Large/unbounded data: `fp-pack/stream`
 
-const handleSubmitPipeline = pipeAsyncSideEffect(
-  tap((e: Event) => e.preventDefault()),
-  prop('currentTarget'),
-  (form) => getFormData(form as HTMLFormElement),
-  validateFieldsOrStop,          // Returns data or SideEffect
-  sanitizeInput,
-  submitToAPI
-);
-
-const handleSubmit = (e: Event) => runPipeResult(handleSubmitPipeline(e));
-```
-
-### Pattern 2: Computing Derived/Reactive Values
-
-**When**: Displaying filtered/sorted/transformed data from state
-**Where to use**: Computed properties, memoized values, derived state
-
-```typescript
-import { pipe, filter, sortBy, map, take } from 'fp-pack';
-
-// GOOD: Create reusable data transformation
-const processUsers = pipe(
-  filter((u: User) => u.status === 'active'),
-  sortBy(u => u.lastLogin),
-  map(u => ({ ...u, displayName: `${u.firstName} ${u.lastName}` })),
-  take(50)
-);
-
-// Use in any framework:
-// React: const processed = useMemo(() => processUsers(users), [users]);
-// Vue: const processed = computed(() => processUsers(users.value));
-// Svelte: $: processed = processUsers($users);
-// Solid: const processed = createMemo(() => processUsers(users()));
-
-// GOOD: Search + filter + pagination
-const searchUsers = (query: string, page: number) =>
-  pipe(
-    filter((u: User) =>
-      u.name.toLowerCase().includes(query.toLowerCase())
-    ),
-    sortBy(u => u.name),
-    chunk(20),              // Paginate
-    (pages) => pages[page] || []
-  );
-
-// React example:
-// const results = useMemo(
-//   () => searchUsers(searchQuery, currentPage)(allUsers),
-//   [searchQuery, currentPage, allUsers]
-// );
-```
-
-### Pattern 3: Async Data Fetching and Processing
-
-**When**: API calls, database queries, file operations
-**Where to use**: Lifecycle hooks, effects, async event handlers
-
-```typescript
-import { pipeAsync, pipeAsyncSideEffect, tap, SideEffect, runPipeResult } from 'fp-pack';
-import { filter, map } from 'fp-pack';
-
-// GOOD: Fetch + transform + update state
-const fetchAndProcessUsers = pipeAsync(
-  async (userId: string) => fetch(`/api/users/${userId}/friends`),
-  async (res) => res.json(),
-  filter((u: User) => u.isActive),
-  map(u => ({ id: u.id, name: u.name, avatar: u.avatar })),
-  (processed) => {
-    setUsers(processed);    // Framework-specific state update
-    return processed;
-  }
-);
-
-// Use in any framework:
-// React: useEffect(() => { fetchAndProcessUsers(id); }, [id]);
-// Vue: watchEffect(() => fetchAndProcessUsers(userId.value));
-// Svelte: $: fetchAndProcessUsers($userId);
-
-// GOOD: Error handling with SideEffect
-const validateResponseOrStop = (users: unknown) => {
-  if (!Array.isArray(users)) {
-    return SideEffect.of(() => {
-      setError('Invalid response');
-      return [];
-    }, 'INVALID_RESPONSE');
-  }
-  return users as User[];
-};
-
-const safeFetchUsersPipeline = pipeAsyncSideEffect(
-  fetchUsers,
-  validateResponseOrStop,
-  filter((u: User) => u.verified),
-  tap((users) => setUsers(users))
-);
-
-const safeFetchUsers = () => runPipeResult(safeFetchUsersPipeline());
-```
-
-### Pattern 4: List/Table Data Processing
-
-**When**: Displaying lists, tables, grids with search/filter/sort
-**Where to use**: Component render logic, computed values
-
-```typescript
-import { pipe, filter, sortBy, groupBy, map } from 'fp-pack';
-
-// GOOD: Complete table data pipeline
-const processTableData = (
-  data: Product[],
-  filters: Filters,
-  sortConfig: SortConfig
-) => pipe(
-  // Apply filters
-  filter((p: Product) => {
-    if (filters.category && p.category !== filters.category) return false;
-    if (filters.minPrice && p.price < filters.minPrice) return false;
-    if (filters.maxPrice && p.price > filters.maxPrice) return false;
-    return true;
-  }),
-  // Apply sorting
-  sortBy(sortConfig.direction === 'asc'
-    ? (p) => p[sortConfig.key]
-    : (p) => -p[sortConfig.key]
-  ),
-  // Add row metadata
-  map((product, index) => ({
-    ...product,
-    rowId: `row-${index}`,
-    isEven: index % 2 === 0
-  }))
-)(data);
-
-// GOOD: Group for categorized display
-const groupProductsByCategory = pipe(
-  groupBy((p: Product) => p.category),
-  (grouped) => Object.entries(grouped).map(([category, products]) => ({
-    category,
-    products,
-    count: products.length,
-    totalValue: products.reduce((sum, p) => sum + p.price, 0)
-  }))
-);
-```
-
-### Pattern 5: Form State Management
-
-**When**: Complex forms with validation and state
-**Where to use**: Form submission, field updates, validation
-
-```typescript
-import { pipe, pipeSideEffect, assoc, pick, mapValues, SideEffect, runPipeResult } from 'fp-pack';
-
-// GOOD: Update nested form state immutably
-const updateField = (fieldName: string, value: any) =>
-  pipe(
-    assoc(fieldName, value),
-    (state) => assoc('touched', { ...state.touched, [fieldName]: true }, state)
-  );
-
-// GOOD: Form submission with validation
-const validateFormOrStop = (data: any) => {
-  const errors = validateFormData(data);
-  return Object.keys(errors).length > 0
-    ? SideEffect.of(() => {
-        setFormErrors(errors);
-        return null;
-      }, 'VALIDATION_ERROR')
-    : data;
-};
-
-const submitFormPipeline = pipeSideEffect(
-  pick(['email', 'password', 'name']),  // Only include relevant fields
-  mapValues((v) => typeof v === 'string' ? v.trim() : v),  // Sanitize
-  validateFormOrStop,
-  submitToAPI
-);
-
-const submitForm = (data: any) => runPipeResult(submitFormPipeline(data));
-
-// GOOD: Multi-step form state
-const validateCurrentStepOrStop = (state: any) => {
-  const errors = validateCurrentStep(state);
-  if (!errors) return state;
-  return SideEffect.of(() => {
-    setStepErrors(errors);
-    return state;
-  }, 'STEP_VALIDATION_ERROR');
-};
-
-const goToNextStepPipeline = pipeSideEffect(
-  validateCurrentStepOrStop,
-  (state) => assoc('currentStep', state.currentStep + 1, state)
-);
-
-const goToNextStep = (state: any) => runPipeResult(goToNextStepPipeline(state));
-```
-
-### Pattern 6: Real-time Data Streams
-
-**When**: WebSocket updates, SSE, real-time data
-**Where to use**: WebSocket handlers, event listeners
-
-```typescript
-import { pipe, filter, map, take } from 'fp-pack';
-
-// GOOD: Process incoming WebSocket messages
-const handleWebSocketMessage = pipe(
-  (event: MessageEvent) => JSON.parse(event.data),
-  filter((msg: Message) => msg.type === 'USER_UPDATE'),
-  map(msg => msg.payload),
-  (update) => {
-    // Update state with new data
-    setUsers(prevUsers =>
-      prevUsers.map(u => u.id === update.id ? { ...u, ...update } : u)
-    );
-  }
-);
-
-// websocket.onmessage = handleWebSocketMessage;
-
-// GOOD: Batch updates with stream
-import { pipe as streamPipe, filter as streamFilter, take as streamTake, toArray } from 'fp-pack/stream';
-import { pipeAsync, runPipeResult } from 'fp-pack';
-
-const processBatchUpdates = async (updates: AsyncIterable<Update>) => {
-  const processed = await streamPipe(
-    streamFilter((u: Update) => u.priority === 'high'),
-    streamTake(100),
-    toArray
-  )(updates);
-
-  batchUpdateUI(processed);
-};
-```
-
-### Pattern 7: Component Props Transformation
-
-**When**: Passing data to child components
-**Where to use**: Component composition, prop drilling
-
-```typescript
-import { pipe, pick, map, merge } from 'fp-pack';
-
-// GOOD: Transform data for child component
-const prepareUserCardProps = pipe(
-  pick(['id', 'name', 'avatar', 'status']),
-  merge({
-    onClick: handleUserClick,
-    className: 'user-card'
-  })
-);
-
-// Usage:
-// const userProps = prepareUserCardProps(user);
-// <UserCard {...userProps} />
-
-// GOOD: Prepare list of component props
-const prepareListItems = pipe(
-  filter((item: Item) => item.visible),
-  map(item => ({
-    key: item.id,
-    ...pick(['title', 'description', 'icon'], item),
-    onClick: () => handleClick(item.id),
-    isActive: item.id === activeId
-  }))
-);
-
-// Usage:
-// {prepareListItems(items).map(props => <ListItem {...props} />)}
-```
-
-### Pattern 8: State Update Reducers
-
-**When**: Complex state updates, global state management
-**Where to use**: Redux/Zustand/Pinia reducers, state update functions
-
-```typescript
-import { pipe, assoc, dissoc, merge, evolve } from 'fp-pack';
-
-// GOOD: Redux-style reducer with fp-pack
-const userReducer = (state: State, action: Action) => {
-  switch (action.type) {
-    case 'ADD_USER':
-      return pipe(
-        prop('users'),
-        append(action.payload),
-        (users) => assoc('users', users, state)
-      )(state);
-
-    case 'UPDATE_USER':
-      return pipe(
-        prop('users'),
-        map((u: User) => u.id === action.payload.id
-          ? merge(u, action.payload.updates)
-          : u
-        ),
-        (users) => assoc('users', users, state)
-      )(state);
-
-    case 'DELETE_USER':
-      return pipe(
-        prop('users'),
-        filter((u: User) => u.id !== action.payload),
-        (users) => assoc('users', users, state)
-      )(state);
-
-    default:
-      return state;
-  }
-};
-
-// GOOD: Using evolve for nested updates
-const updateNestedState = evolve({
-  user: evolve({
-    profile: merge({ verified: true }),
-    settings: assoc('notifications', false)
-  }),
-  lastUpdated: () => new Date()
-});
-```
-
-### Pattern 9: Optimistic Updates
-
-**When**: UI updates before server confirmation
-**Where to use**: Create/update/delete operations
-
-```typescript
-import { pipe, pipeAsync, append, filter } from 'fp-pack';
-
-// GOOD: Optimistic create with rollback
-const createItemOptimistically = (newItem: Item) => {
-  const tempId = `temp-${Date.now()}`;
-  const optimisticItem = { ...newItem, id: tempId, pending: true };
-
-  // Immediately update UI
-  setItems(pipe(append(optimisticItem)));
-
-  // Then persist
-  return pipeAsync(
-    async () => api.createItem(newItem),
-    (savedItem) => {
-      // Replace temp with real item
-      setItems(
-        pipe(
-          filter((item: Item) => item.id !== tempId),
-          append(savedItem)
-        )
-      );
-      return savedItem;
-    }
-  )().catch((error) => {
-    // Rollback on error
-    setItems(pipe(filter((item: Item) => item.id !== tempId)));
-    throw error;
-  });
-};
-```
-
-### Pattern 10: URL/Query Parameter Handling
-
-**When**: Syncing UI state with URL
-**Where to use**: Routing, search parameters, filters
-
-```typescript
-import { pipe, pick, mapValues, merge } from 'fp-pack';
-
-// GOOD: Parse query params to state
-const parseQueryParams = pipe(
-  (search: string) => new URLSearchParams(search),
-  (params) => ({
-    page: Number(params.get('page')) || 1,
-    query: params.get('q') || '',
-    category: params.get('category') || 'all',
-    sort: params.get('sort') || 'date'
-  })
-);
-
-// GOOD: Convert state to query params
-const stateToQueryParams = pipe(
-  pick(['page', 'query', 'category', 'sort']),
-  (state) => {
-    const params = new URLSearchParams();
-    Object.entries(state).forEach(([key, value]) => {
-      if (value) params.set(key, String(value));
-    });
-    return params.toString();
-  }
-);
-
-// Usage in framework router:
-// const filters = parseQueryParams(location.search);
-// navigate(`/products?${stateToQueryParams(currentState)}`);
-```
-
-### Pattern 11: Infinite Scroll / Virtual Lists
-
-**When**: Large lists with lazy loading, infinite scroll, virtual rendering
-**Where to use**: Scroll handlers, pagination, large dataset rendering
-
-```typescript
-import { pipe, pipeAsyncSideEffect, when, tap, runPipeResult } from 'fp-pack';
-import { pipe as streamPipe, filter as streamFilter, take as streamTake, toArray } from 'fp-pack/stream';
-
-// GOOD: Infinite scroll with pipe - all logic inside
-const handleScroll = pipe(
-  (e: Event) => e.target as HTMLElement,
-  (el) => ({
-    scrollTop: el.scrollTop,
-    scrollHeight: el.scrollHeight,
-    clientHeight: el.clientHeight,
-    hasMore,
-    isLoading
-  }),
-  // Only load if near bottom, has more data, and not currently loading
-  when(
-    ({ scrollTop, scrollHeight, clientHeight, hasMore, isLoading }) =>
-      scrollTop + clientHeight >= scrollHeight - 100 && hasMore && !isLoading,
-    tap(() => loadNextPage())
-  )
-);
-
-// GOOD: Load next page with stream processing
-const loadNextPagePipeline = pipeAsyncSideEffect(
-  async () => {
-    setIsLoading(true);
-    return fetchItemsFromAPI(currentPage);
-  },
-  // Use stream for lazy processing
-  (items) => streamPipe(
-    streamFilter((item: Item) => item.visible),
-    streamTake(pageSize),
-    toArray
-  )(items),
-  tap((newItems) => setItems(prev => [...prev, ...newItems])),
-  tap(() => setCurrentPage(prev => prev + 1)),
-  tap(() => setIsLoading(false))
-);
-
-const loadNextPage = () => runPipeResult(loadNextPagePipeline());
-
-// GOOD: Virtual scroll - calculate visible range in pipe
-const getVisibleItems = pipe(
-  (scrollTop: number) => ({
-    itemHeight: 50,
-    viewportHeight: 600,
-    bufferSize: 5,
-    scrollTop
-  }),
-  ({ itemHeight, viewportHeight, bufferSize, scrollTop }) => ({
-    startIndex: Math.floor(scrollTop / itemHeight),
-    endIndex: Math.ceil((scrollTop + viewportHeight) / itemHeight),
-    bufferSize
-  }),
-  ({ startIndex, endIndex, bufferSize }) => ({
-    start: Math.max(0, startIndex - bufferSize),
-    end: endIndex + bufferSize
-  }),
-  ({ start, end }) => allItems.slice(start, end)
-);
-
-// GOOD: Lazy load with intersection observer
-const handleIntersection = pipe(
-  (entries: IntersectionObserverEntry[]) => entries[0],
-  when(
-    (entry) => entry.isIntersecting && hasMoreItems && !isLoadingMore,
-    tap(() => loadMoreItems())
-  )
-);
-```
-
-### Pattern 12: Conditional State Updates
-
-**When**: State updates that depend on conditions
-**Where to use**: Any state update with business logic
-
-```typescript
-import { pipe, ifElse, when, cond, tap, assoc, prop, append, map, filter, merge } from 'fp-pack';
-
-// GOOD: Use ifElse instead of if/else
-const toggleUserStatus = pipe(
-  ifElse(
-    (user: User) => user.status === 'active',
-    assoc('status', 'inactive'),
-    assoc('status', 'active')
-  ),
-  tap((updatedUser) => setUser(updatedUser))
-);
-
-// GOOD: Use ifElse for conditional side effects
-const saveIfValid = pipe(
-  validateForm,
-  ifElse(
-    (result) => result.isValid,
-    pipe(
-      tap((data) => saveToAPI(data)),
-      tap(() => showSuccessMessage())
-    ),
-    tap((result) => setErrors(result.errors))
-  )
-);
-
-// GOOD: Use cond for multi-branch logic (instead of switch/if-else chain)
-const processUserAction = pipe(
-  prop('action'),
-  cond([
-    [
-      (action) => action.type === 'CREATE',
-      pipe(
-        prop('payload'),
-        (user) => append(user),
-        tap((users) => setUsers(users))
-      )
-    ],
-    [
-      (action) => action.type === 'UPDATE',
-      pipe(
-        prop('payload'),
-        ({ id, updates }) => map((u: User) => u.id === id ? merge(u, updates) : u),
-        tap((users) => setUsers(users))
-      )
-    ],
-    [
-      (action) => action.type === 'DELETE',
-      pipe(
-        prop('payload'),
-        (id) => filter((u: User) => u.id !== id),
-        tap((users) => setUsers(users))
-      )
-    ],
-    [
-      () => true, // default case
-      tap(() => console.warn('Unknown action'))
-    ]
-  ])
-);
-
-// GOOD: Complex state update with all logic in pipe
-const updateCartItem = (itemId: string, quantity: number) => pipe(
-  // Get current cart
-  (cart) => cart.items,
-  // Find and update item
-  map((item: CartItem) =>
-    ifElse(
-      () => item.id === itemId,
-      pipe(
-        assoc('quantity', quantity),
-        when(
-          (updated) => updated.quantity <= 0,
-          () => null  // Mark for removal
-        )
-      ),
-      () => item
-    )(item)
-  ),
-  // Remove null items (quantity <= 0)
-  filter((item) => item !== null),
-  // Update state
-  tap((items) => setCart({ items })),
-  // Show notification
-  tap((items) => {
-    const item = items.find(i => i.id === itemId);
-    if (item) showNotification(`Updated ${item.name}`);
-    else showNotification('Item removed from cart');
-  })
-);
-```
-
-## Library Integration Quick Reference
-
-This section shows how to integrate fp-pack with popular UI libraries. All examples keep logic **inside pipe chains** using fp-pack control flow functions.
-
-### React Ecosystem
-
-#### React Query / TanStack Query
-
-```typescript
-import { useQuery, useMutation } from '@tanstack/react-query';
-import { pipe, filter, sortBy, map, tap, when } from 'fp-pack';
-
-// GOOD: Transform data in select using pipe
-const { data: activeUsers } = useQuery({
-  queryKey: ['users'],
-  queryFn: fetchUsers,
-  select: pipe(
-    filter((u: User) => u.status === 'active'),
-    sortBy((u) => u.name),
-    map((u) => ({ id: u.id, name: u.name, email: u.email }))
-  )
-});
-
-// GOOD: Handle mutations with pipe
-const mutation = useMutation({
-  mutationFn: createUser,
-  onSuccess: pipe(
-    tap((newUser) => queryClient.invalidateQueries(['users'])),
-    when(
-      (user) => user.isPremium,
-      tap(() => showPremiumWelcome())
-    ),
-    tap(() => navigate('/dashboard'))
-  )
-});
-
-// GOOD: Optimistic updates in pipe
-const updateMutation = useMutation({
-  mutationFn: updateUser,
-  onMutate: pipe(
-    tap(async (newUser) => {
-      await queryClient.cancelQueries(['users']);
-      const previous = queryClient.getQueryData(['users']);
-      queryClient.setQueryData(['users'], pipe(
-        map((u: User) => u.id === newUser.id ? merge(u, newUser) : u)
-      ));
-      return { previous };
-    })
-  ),
-  onError: pipe(
-    tap((err, variables, context) => {
-      if (context?.previous) {
-        queryClient.setQueryData(['users'], context.previous);
-      }
-    })
-  )
-});
-```
-
-#### Zustand
-
-```typescript
-import create from 'zustand';
-import { pipe, append, filter, map, merge, ifElse, when, tap, prop, sortBy, assoc } from 'fp-pack';
-
-// GOOD: All actions use pipe
-const useStore = create((set, get) => ({
-  users: [],
-
-  addUser: pipe(
-    (user: User) => user,
-    when(
-      (user) => !get().users.some(u => u.id === user.id),
-      tap((user) => set(pipe(
-        prop('users'),
-        append(user),
-        sortBy((u: User) => u.name),
-        (users) => ({ users })
-      )(get())))
-    )
-  ),
-
-  updateUser: (id: string, updates: Partial<User>) => set(pipe(
-    prop('users'),
-    map((u: User) =>
-      ifElse(
-        () => u.id === id,
-        merge(u, updates),
-        () => u
-      )(u)
-    ),
-    (users) => ({ users })
-  )(get())),
-
-  deleteUser: (id: string) => set(pipe(
-    prop('users'),
-    filter((u: User) => u.id !== id),
-    (users) => ({ users })
-  )(get())),
-
-  toggleUserStatus: (id: string) => set(pipe(
-    prop('users'),
-    map((u: User) => u.id === id
-      ? pipe(
-          ifElse(
-            (user) => user.status === 'active',
-            assoc('status', 'inactive'),
-            assoc('status', 'active')
-          )
-        )(u)
-      : u
-    ),
-    (users) => ({ users })
-  )(get()))
-}));
-```
-
-#### Redux Toolkit
-
-```typescript
-import { createSlice } from '@reduxjs/toolkit';
-import { pipe, append, filter, map, merge, sortBy, cond, assoc } from 'fp-pack';
-
-// GOOD: Reducers with pipe - no manual mutations
-const userSlice = createSlice({
-  name: 'users',
-  initialState: { list: [], loading: false },
-  reducers: {
-    addUser: (state, action) => {
-      state.list = pipe(
-        append(action.payload),
-        sortBy((u: User) => u.createdAt)
-      )(state.list);
-    },
-
-    updateUser: (state, action) => {
-      state.list = pipe(
-        map((u: User) =>
-          u.id === action.payload.id
-            ? merge(u, action.payload.updates)
-            : u
-        )
-      )(state.list);
-    },
-
-    deleteUser: (state, action) => {
-      state.list = pipe(
-        filter((u: User) => u.id !== action.payload)
-      )(state.list);
-    },
-
-    // Complex update with cond
-    processAction: (state, action) => {
-      state.list = pipe(
-        cond([
-          [
-            () => action.type === 'BULK_ACTIVATE',
-            map((u: User) => assoc('status', 'active', u))
-          ],
-          [
-            () => action.type === 'BULK_DELETE',
-            filter((u: User) => !action.payload.ids.includes(u.id))
-          ],
-          [
-            () => true,
-            (users) => users  // no change
-          ]
-        ])
-      )(state.list);
-    }
-  }
-});
-```
-
-#### React Hook Form
-
-```typescript
-import { useForm } from 'react-hook-form';
-import { pipe, pipeSideEffect, pipeAsyncSideEffect, pick, mapValues, trim, when, tap, SideEffect, runPipeResult } from 'fp-pack';
-
-// GOOD: Validation with pipeSideEffect
-const validateFormDataPipeline = pipeSideEffect(
-  pick(['email', 'password', 'name']),
-  mapValues((v) => typeof v === 'string' ? trim(v) : v),
-  (data) => {
-    const errors: any = {};
-    if (!data.email?.includes('@')) errors.email = 'Invalid email';
-    if ((data.password?.length || 0) < 8) errors.password = 'Too short';
-    return Object.keys(errors).length > 0
-      ? SideEffect.of(() => ({ values: {}, errors }), 'VALIDATION_ERROR')
-      : { values: data, errors: {} };
-  }
-);
-
-const validateFormData = (values: any) => runPipeResult(validateFormDataPipeline(values));
-
-const { register, handleSubmit } = useForm({
-  resolver: (values) => validateFormData(values)
-});
-
-// GOOD: Submit handler with pipeAsyncSideEffect
-const onSubmitPipeline = pipeAsyncSideEffect(
-  validateFormData,
-  when(
-    (result) => Object.keys(result.errors).length === 0,
-    pipe(
-      prop('values'),
-      submitToAPI,
-      tap(() => navigate('/success'))
-    )
-  )
-);
-
-const onSubmit = (data: any) => runPipeResult(onSubmitPipeline(data));
-```
+---
 
-### Vue Ecosystem
-
-#### Pinia
-
-```typescript
-import { defineStore } from 'pinia';
-import { pipe, append, filter, map, merge, sortBy, when, tap } from 'fp-pack';
-
-// GOOD: All actions use pipe
-export const useUserStore = defineStore('user', {
-  state: () => ({ users: [], loading: false }),
-
-  actions: {
-    addUser(user: User) {
-      this.users = pipe(
-        append(user),
-        sortBy((u: User) => u.name),
-        when(
-          (users) => users.length > 100,
-          tap(() => this.showWarning('Many users'))
-        )
-      )(this.users);
-    },
-
-    updateUser(id: string, updates: Partial<User>) {
-      this.users = pipe(
-        map((u: User) => u.id === id ? merge(u, updates) : u)
-      )(this.users);
-    },
-
-    deleteUser(id: string) {
-      this.users = pipe(
-        filter((u: User) => u.id !== id),
-        tap((users) => {
-          if (users.length === 0) this.showEmptyState = true;
-        })
-      )(this.users);
-    }
-  }
-});
-```
+## Anti-Patterns (Avoid)
 
-#### VueUse
-
-```typescript
-import { useFetch } from '@vueuse/core';
-import { pipe, filter, map, sortBy, tap } from 'fp-pack';
-
-// GOOD: Transform response with pipe
-const { data } = useFetch('/api/users', {
-  afterFetch: pipe(
-    prop('data'),
-    filter((u: User) => u.verified),
-    sortBy((u) => u.name),
-    map((u) => ({ id: u.id, name: u.name })),
-    tap((users) => console.log(`Loaded ${users.length} users`))
-  )
-}).json();
-
-// GOOD: Refetch with condition in pipe
-const { execute } = useFetch('/api/users');
-
-const refreshIfNeeded = pipe(
-  (lastUpdate: Date) => Date.now() - lastUpdate.getTime(),
-  when(
-    (diff) => diff > 5 * 60 * 1000, // 5 minutes
-    tap(() => execute())
-  )
-);
-```
+- Imperative loops when a pipeline is clearer
+- Mutating inputs (prefer immutable transforms)
+- Chaining `Array.prototype.*` in complex transforms (prefer `pipe`)
+- Calling `runPipeResult` inside a SideEffect-aware pipeline
+- Adding “classic monads” to emulate Option/Either instead of using `SideEffect` pipes when necessary
 
-### State Management Patterns
-
-All state management libraries benefit from fp-pack's immutable update patterns:
-
-```typescript
-// GOOD: Generic state update pattern (works with any library)
-const updateState = <T>(
-  state: T,
-  path: string[],
-  updater: (value: any) => any
-) => pipe(
-  pathOr(null, path),
-  updater,
-  (newValue) => assocPath(path, newValue, state)
-)(state);
-
-// Usage in any framework:
-// Redux: return updateState(state, ['users', 0, 'name'], toUpper);
-// Zustand: set(updateState(get(), ['users', 0, 'name'], toUpper));
-// Pinia: this.$state = updateState(this.$state, ['users', 0, 'name'], toUpper);
-```
+---
 
-## Framework-Specific Notes
+## Quick Signature Lookup (When Unsure)
 
-While the patterns above are framework-agnostic, here's where to apply them:
+Prefer local types first:
+- Main types: `dist/index.d.ts`
+- Stream types: `dist/stream/index.d.ts`
 
-### Reactive/Computed Values
-- **React**: `useMemo(() => pipe(...)(data), [data])`
-- **Vue**: `computed(() => pipe(...)(data.value))`
-- **Svelte**: `$: result = pipe(...)(data)`
-- **Solid**: `createMemo(() => pipe(...)(data()))`
+If you’re in a consumer project:
+- `node_modules/fp-pack/dist/index.d.ts`
+- `node_modules/fp-pack/dist/stream/index.d.ts`
 
-### Event Handlers
-- **React**: `<button onClick={pipe(...)}>` or `const handler = pipe(...)`
-- **Vue**: `<button @click="pipe(...)">` or `const handler = pipe(...)`
-- **Svelte**: `<button on:click={pipe(...)}>` or `const handler = pipe(...)`
+---
 
-### Side Effects (API calls, subscriptions)
-- **React**: `useEffect(() => { pipeAsync(...)() }, [deps])`
-- **Vue**: `watchEffect(() => pipeAsync(...)())`
-- **Svelte**: `onMount(() => pipeAsync(...)())`
+## Writing New Helpers (If Needed)
 
-### State Updates
-- **React**: `setState(pipe(...)(currentState))`
-- **Vue**: `state.value = pipe(...)(state.value)`
-- **Svelte**: `$state = pipe(...)($state)`
+If you add your own utilities that should compose well:
+- Keep them **unary** for pipelines (or return a unary function via currying).
+- Prefer **data-last** argument order.
+- For generic/overloaded helpers, consider providing an explicit type alias to preserve inference in TypeScript.
 
-All patterns use the same fp-pack functions - only the framework's state/reactive wrapper changes.
+---
 
 ## Summary
 
-As an AI coding assistant working with fp-pack:
-
-1. **Default to `pipe`** for all data transformations
-2. **Switch to `pipeAsync`** when async operations are involved (use `pipeAsyncSideEffect` if SideEffect is in the flow)
-3. **Use `stream/*`** for lazy, memory-efficient processing
-4. **Handle errors with `SideEffect`** in `pipeSideEffect`/`pipeAsyncSideEffect`, not try-catch
-5. **Avoid imperative loops** - use fp-pack's declarative functions
-6. **Never suggest monads** - use SideEffect pattern instead
-7. **Keep code declarative** - describe what, not how
-8. **All logic inside pipe** - never break out of pipe chains for conditionals or loops
-9. **Use control flow functions** - `when`, `unless`, `ifElse`, `cond` instead of if/else/switch
-10. **Call `runPipeResult` OUTSIDE pipelines** - `runPipeResult` / `matchSideEffect` must be called outside `pipeSideEffect`/`pipeAsyncSideEffect` for proper type safety
-11. **Use `isSideEffect` for type narrowing** - get precise types in both success and error branches
-12. **Apply use-case patterns** - recognize scenarios (form handling, list processing, etc.) and apply appropriate fp-pack patterns
-13. **Framework-agnostic core** - write fp-pack logic independent of UI framework, only wrap at the boundaries
-14. **Library integration** - use pipe in select/resolver/action functions of popular libraries (React Query, Zustand, Pinia, etc.)
-
-### Key Principles
-
-**✅ DO: Keep everything in pipe**
-```typescript
-// GOOD: All logic inside pipe
-const handleSubmitPipeline = pipeAsyncSideEffect(
-  getFormData,
-  validateFields,
-  when(isValid, submitToAPI),
-  unless(isValid, showErrors)
-);
-
-const handleSubmit = (form: HTMLFormElement) => runPipeResult(handleSubmitPipeline(form));
-```
-
-**❌ DON'T: Break out of pipe for conditionals**
-```typescript
-// BAD: Breaking pipe for if/else
-const handleSubmit = pipe(
-  getFormData,
-  validateFields
-);
-const result = handleSubmit(form);
-if (result.isValid) {  // ❌ Outside pipe
-  submitToAPI(result);
-} else {
-  showErrors(result.errors);
-}
-```
-
-**✅ DO: Use when/cond/ifElse for branching**
-```typescript
-// GOOD: Branching inside pipe
-const processAction = pipe(
-  cond([
-    [(action) => action.type === 'CREATE', handleCreate],
-    [(action) => action.type === 'UPDATE', handleUpdate],
-    [(action) => action.type === 'DELETE', handleDelete],
-    [() => true, handleDefault]
-  ])
-);
-```
-
-**❌ DON'T: Use switch/if-else chains**
-```typescript
-// BAD: Imperative branching
-const processAction = (action) => {
-  switch (action.type) {  // ❌ Imperative
-    case 'CREATE': return handleCreate(action);
-    case 'UPDATE': return handleUpdate(action);
-    case 'DELETE': return handleDelete(action);
-    default: return handleDefault(action);
-  }
-};
-```
-
-Your goal is to write clean, readable, functional code that leverages fp-pack's full potential in real-world UI applications.
+Default to `pipe` / `pipeAsync`, keep helpers data-last and unary, switch to `stream/*` when laziness matters, and reserve SideEffect-aware pipelines for true early-exit flows. Use `isSideEffect` for precise narrowing and call `runPipeResult` only at the boundary (with generics if inference widens to `any`).